def __init__(self, weights_path):
self.model = SSD('test')
self.model.cuda().eval()
state = torch.load(weights_path, map_location=lambda storage, loc: storage)
state = {key: value.float() for key, value in state.items()}
self.model.load_state_dict(state)
self.transform = GeneralizedRCNNTransform(DETECTOR_MIN_SIZE, DETECTOR_MAX_SIZE, DETECTOR_MEAN, DETECTOR_STD)
self.transform.eval()
def _init_test_generalized_rcnn_transform(self):
min_size = 100
max_size = 200
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
return transform
def test_not_float_normalize(self):
transform = GeneralizedRCNNTransform(300, 500, torch.zeros(3),
torch.ones(3))
image = [torch.randint(0, 255, (3, 200, 300), dtype=torch.uint8)]
targets = [{"boxes": torch.rand(3, 4)}]
with pytest.raises(TypeError):
out = transform(image, targets) # noqa: F841
def __init__(self_module):
super(TransformModule, self_module).__init__()
min_size = 800
max_size = 1333
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
self_module.transform = GeneralizedRCNNTransform(
min_size, max_size, image_mean, image_std)
def test_transform_copy_targets(self):
transform = GeneralizedRCNNTransform(300, 500, torch.zeros(3), torch.ones(3))
image = [torch.rand(3, 200, 300), torch.rand(3, 200, 200)]
targets = [{'boxes': torch.rand(3, 4)}, {'boxes': torch.rand(2, 4)}]
targets_copy = copy.deepcopy(targets)
out = transform(image, targets) # noqa: F841
self.assertTrue(torch.equal(targets[0]['boxes'], targets_copy[0]['boxes']))
self.assertTrue(torch.equal(targets[1]['boxes'], targets_copy[1]['boxes']))
def test_transform_copy_targets(self):
transform = GeneralizedRCNNTransform(300, 500, torch.zeros(3), torch.ones(3))
image = [torch.rand(3, 200, 300), torch.rand(3, 200, 200)]
targets = [{"boxes": torch.rand(3, 4)}, {"boxes": torch.rand(2, 4)}]
targets_copy = copy.deepcopy(targets)
out = transform(image, targets) # noqa: F841
assert_equal(targets[0]["boxes"], targets_copy[0]["boxes"])
assert_equal(targets[1]["boxes"], targets_copy[1]["boxes"])
def __init__(self, root, img_transform, extra_info=True):
self.root = root
self.img_transform = img_transform
self.extra_info = extra_info
self.anno = pd.read_csv(
os.path.join(self.root, 'annotation.csv'))
self.target_transform = GeneralizedRCNNTransform(
400, 400, [0., 0., 0.], [1., 1., 1.])
class Detector(object):
def __init__(self, weights_path):
self.model = SSD('test')
self.model.cuda().eval()
state = torch.load(weights_path,
map_location=lambda storage, loc: storage)
state = {key: value.float() for key, value in state.items()}
self.model.load_state_dict(state)
self.transform = GeneralizedRCNNTransform(DETECTOR_MIN_SIZE,
DETECTOR_MAX_SIZE,
DETECTOR_MEAN, DETECTOR_STD)
self.transform.eval()
def detect(self, images):
images = torch.stack(
[torch.from_numpy(image).cuda() for image in images])
images = images.transpose(1, 3).transpose(2, 3).float()
original_image_sizes = [img.shape[-2:] for img in images]
images, _ = self.transform(images, None)
with torch.no_grad():
detections_batch = self.model(images.tensors).cpu().numpy()
result = []
for detections, image_size in zip(detections_batch,
images.image_sizes):
scores = detections[1, :, 0]
keep_idxs = scores > DETECTOR_THRESHOLD
detections = detections[1, keep_idxs, :]
detections = detections[:, [1, 2, 3, 4, 0]]
detections[:, 0] *= image_size[1]
detections[:, 1] *= image_size[0]
detections[:, 2] *= image_size[1]
detections[:, 3] *= image_size[0]
result.append({
'scores': torch.from_numpy(detections[:, 4]),
'boxes': torch.from_numpy(detections[:, :4])
})
result = self.transform.postprocess(result, images.image_sizes,
original_image_sizes)
return result
def __init__(self, backbone, num_classes,
min_size=800, max_size=1333,
image_mean=None, image_std=None,
anchor_generator=None, head=None,
proposal_matcher=None,
score_thresh=0.05,
nms_thresh=0.5,
detections_per_img=300,
fg_iou_thresh=0.5, bg_iou_thresh=0.4,
topk_candidates=1000):
super(RetinaNet, self).__init__()
if not hasattr(backbone, "out_channels"):
raise ValueError("backbone should contain an attribute out_channels specifying the number of output channels "
"assumed be the samefor all the levels")
self.backbone = backbone
assert isinstance(anchor_generator, (AnchorGenerator, type(None)))
if anchor_generator is None:
anchor_sizes = tuple((x, int(x * 2 ** (1.0 / 3)), int(x * 2 ** (2.0 / 3))) for x in [32, 64, 128, 256, 512])
aspect_ratios = ((0.5, 1.0, 2.0),) * len(anchor_sizes)
anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
self.anchor_generator = anchor_generator
if head is None:
head = RetinaNetHead(backbone.out_channels, anchor_generator.num_anchors_per_location()[0], num_classes)
self.head = head
if proposal_matcher is None:
proposal_matcher = det_utils.Matcher(
fg_iou_thresh,
bg_iou_thresh,
allow_low_quality_matches = True,
)
self.proposal_matcher = proposal_matcher
self.box_coder = det_utils.BoxCoder(weights=(1.0, 1.0, 1.0, 1.0))
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
self.transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
self.score_thresh = score_thresh
self.nms_thresh = nms_thresh
self.detections_per_img = detections_per_img
self.topk_candidates = topk_candidates
self.has_warned = False
def get_features_for_projection(model, imagePath, device):
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
# these transform parameters are from source code of Mask R CNN
transform = GeneralizedRCNNTransform(min_size=800, max_size=1333, image_mean=image_mean, image_std=image_std)
image = Image.open(imagePath)
image_tensor = TF.to_tensor(image)
# let it be in list (can be multiple)
# TODO make it multiple
images = [image_tensor]
images, _ = transform(images)
features = model.backbone(images.tensors.to(device))
features_to_be_projected = features['pool']
return features_to_be_projected
def get_features_for_projection_multi(model, imagePaths, device):
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
# these transform parameters are from source code of Mask R CNN
transform = GeneralizedRCNNTransform(min_size=800, max_size=1333, image_mean=image_mean, image_std=image_std)
images = [Image.open(imagePath) for imagePath in imagePaths]
image_tensors = [TF.to_tensor(image) for image in images]
# let it be in list (can be multiple)
results = []
with torch.no_grad():
for tensor in image_tensors:
images, _ = transform([tensor])
features = model.backbone(images.tensors.to(device))
features_to_be_projected = features['pool']
results.append(features_to_be_projected[0])
return results
def __init__(self,
backbone,
rpn,
roi_heads,
mask_net,
transform,
input_img_num=6,
depth_estimator_path='_depth_net.pth'):
super(GeneralizedRCNN, self).__init__()
self.transform = transform
self.backbone = UNet(4, 1)
self.backbone_ = UNet(4, 64)
self.input_img_num = input_img_num
self.rpn = rpn
self.roi_heads = roi_heads
self.mask_net = UnetMask(6, 1)
self.backbone_out_channels = 64
self.depth_estimator_path = depth_estimator_path
self.depth_estimator = VggDepthEstimator()
self.depth_estimator.load_state_dict(
torch.load(self.depth_estimator_path))
self.depth_resize = nn.Upsample(size=(400, 400),
mode='bilinear',
align_corners=True)
self.img_transform = transforms.Compose([
transforms.ToPILImage(),
transforms.Resize((400, 400)),
transforms.ToTensor(),
transforms.Normalize(MEAN, STD)
])
self.target_transform = GeneralizedRCNNTransform(
400, 400, [0., 0., 0.], [1., 1., 1.])
self.mask_transform = transforms.Compose([
transforms.ToPILImage(),
transforms.Resize((800, 800)),
transforms.ToTensor()
])
self.depth_transform = transforms.Compose([
transforms.ToPILImage(),
transforms.Resize((128, 416)),
transforms.ToTensor(),
transforms.Normalize(MEAN, STD)
])
def __init__(self, backbone, rpn, roi_heads, transform, input_img_num=6):
super(DetectionGeneralizedRCNN, self).__init__()
self.transform = transform
self.backbone = backbone
self.input_img_num = input_img_num
self.rpn = rpn
self.roi_heads = roi_heads
self.backbone_out_channels = backbone.out_channels
self.img_transform = transforms.Compose([
transforms.ToPILImage(),
transforms.Resize((400, 400)),
transforms.ToTensor(),
transforms.Normalize(MEAN, STD)
])
self.target_transform = GeneralizedRCNNTransform(
400, 400, [0., 0., 0.], [1., 1., 1.])
def store_features_for_projection_multi(imagePaths, model_maskrcnn, outputs,
sceneid):
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
# these transform parameters are from source code of Mask R CNN
transform = GeneralizedRCNNTransform(min_size=800,
max_size=1333,
image_mean=image_mean,
image_std=image_std)
images = [Image.open(imagePath) for imagePath in imagePaths]
image_tensors = [TF.to_tensor(image) for image in images]
# let it be in list (can be multiple)
images, _ = transform(image_tensors)
with torch.no_grad():
body = model_maskrcnn.body
body = body
output = body(images.tensors)
torch.save(output, outputs + sceneid + ".fea")
def __init__(
self,
backbone,
num_classes=2,
num_pids=5532,
num_cq_size=5000,
# transform parameters
min_size=900,
max_size=1500,
image_mean=None,
image_std=None,
# Anchor settings:
anchor_scales=None,
anchor_ratios=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=12000,
rpn_pre_nms_top_n_test=6000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=300,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
# Box parameters
rcnn_bbox_bn=True,
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.4,
box_detections_per_img=300,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.1,
box_batch_size_per_image=128,
box_positive_fraction=0.5,
bbox_reg_weights=None,
# ReID parameters
feat_head=None,
reid_head=None,
reid_loss=None):
if rpn_anchor_generator is None:
anchor_sizes = ((32, 64, 128, 256, 512), )
aspect_ratios = ((0.5, 1.0, 2.0), )
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
rpn_head = RPNHead(backbone.out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator, rpn_head,
rpn_fg_iou_thresh, rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction, rpn_pre_nms_top_n,
rpn_post_nms_top_n, rpn_nms_thresh)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(featmap_names=['feat2rpn'],
output_size=[14, 14],
sampling_ratio=2)
if box_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 2048
box_head = GAP_BOX_HEAD(resolution, feat_head, representation_size)
if box_predictor is None:
representation_size = 2048
box_predictor = FastRCNNPredictor(representation_size,
num_classes,
RCNN_bbox_bn=False)
if reid_head is None:
reid_head = REID_HEAD(box_head.out_dims, 256)
if reid_loss is None:
reid_loss = OIMLoss(256, num_pids, num_cq_size, 0.5, 30)
roi_heads = OIM_ROI_HEAD(
reid_head,
reid_loss,
# box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(FasterRCNN_OIM, self).__init__(backbone, rpn, roi_heads,
transform)
def evaluate_yolo_2017(model, data_loader, device):
n_threads = torch.get_num_threads()
# FIXME remove this and make paste_masks_in_image run on the GPU
torch.set_num_threads(1)
cpu_device = torch.device("cpu")
model.eval()
metric_logger = utils.MetricLogger(delimiter=" ")
header = 'Test:'
coco = get_coco_api_from_dataset(data_loader.dataset)
iou_types = _get_iou_types(model)
coco_evaluator = CocoEvaluator(coco, iou_types)
transform = GeneralizedRCNNTransform(416, 416, [0, 0, 0], [1, 1, 1])
transform.eval()
for image, targets in metric_logger.log_every(data_loader, 100, header):
image = list(img.to(device) for img in image)
original_image_sizes = [img.shape[-2:] for img in image]
targets = [{k: v.to(device) for k, v in t.items()} for t in targets]
torch.cuda.synchronize()
model_time = time.time()
transformed_img = transform(image)
transformed_shape = transformed_img[0].tensors.shape[-2:]
inf_out, _ = model(transformed_img[0].tensors)
# Run NMS
output = non_max_suppression(inf_out, conf_thres=0.001, iou_thres=0.6)
# Statistics per image
predictions = []
for si, pred in enumerate(output):
prediction = {'boxes': [], 'labels': [], 'scores': []}
if pred is None:
continue
# Append to text file
# with open('test.txt', 'a') as file:
# [file.write('%11.5g' * 7 % tuple(x) + '\n') for x in pred]
# Clip boxes to image bounds
clip_coords(pred, transformed_shape)
# Append to pycocotools JSON dictionary
# [{"image_id": 42, "category_id": 18, "bbox": [258.15, 41.29, 348.26, 243.78], "score": 0.236}, ...
image_id = int(targets[si]['image_id'])
box = pred[:, :4].clone() # xyxy
# scale_coords(transformed_shape, box, shapes[si][0], shapes[si][1]) # to original shape
# box = xyxy2xywh(box) # xywh
# box[:, :2] -= box[:, 2:] / 2 # xy center to top-left corner
for di, d in enumerate(pred):
box_T = [floatn(x, 3) for x in box[di]]
label = coco91class[int(d[5])]
score = floatn(d[4], 5)
prediction['boxes'].append(box_T)
prediction['labels'].append(label)
prediction['scores'].append(score)
prediction['boxes'] = torch.tensor(prediction['boxes'])
prediction['labels'] = torch.tensor(prediction['labels'])
prediction['scores'] = torch.tensor(prediction['scores'])
predictions.append(prediction)
outputs = transform.postprocess(predictions,
transformed_img[0].image_sizes,
original_image_sizes)
outputs = [{k: v.to(cpu_device)
for k, v in t.items()} for t in predictions]
model_time = time.time() - model_time
res = {
target["image_id"].item(): output
for target, output in zip(targets, outputs)
}
evaluator_time = time.time()
coco_evaluator.update(res)
evaluator_time = time.time() - evaluator_time
metric_logger.update(model_time=model_time,
evaluator_time=evaluator_time)
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print("Averaged stats:", metric_logger)
coco_evaluator.synchronize_between_processes()
# accumulate predictions from all images
coco_evaluator.accumulate()
coco_evaluator.summarize()
torch.set_num_threads(n_threads)
return coco_evaluator
def __init__(
self,
backbone,
num_classes=None,
# transform parameters
min_size=800,
max_size=800,
image_mean=None,
image_std=None,
# RPN parameters
anchor_generator=None,
# Box parameters
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.5,
box_detections_per_img=100,
box_fg_iou_thresh=0.15,
box_bg_iou_thresh=0.15,
box_batch_size_per_image=50,
box_positive_fraction=0.5,
bbox_reg_weights=None):
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if anchor_generator is None:
# anchor size per every feature map level
anchor_sizes = ((16, ), (32, ), (64, ), (128, ), (210, ), (320, ))
aspect_ratios = ((0.5, 1.0, 2.0), ) * len(anchor_sizes)
anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
num_anchors_per_location = anchor_generator.num_anchors_per_location(
)[0]
ssd_predictor = SSDPredictor(out_channels, num_classes,
num_anchors_per_location)
ssd_head = SSDHead(
# Box
ssd_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights)
bbox_reg_weights = ssd_head.bbox_reg_weight
detection_filter = DetectionNmsPostprocessor(box_score_thresh,
box_nms_thresh,
bbox_reg_weights,
box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(SSD, self).__init__(transform, backbone, anchor_generator,
ssd_head, detection_filter, num_classes)
def __init__(self, cfg):
super(SeqNet, self).__init__()
backbone, box_head = build_resnet(name="resnet50", pretrained=True)
anchor_generator = AnchorGenerator(sizes=((32, 64, 128, 256, 512), ),
aspect_ratios=((0.5, 1.0, 2.0), ))
head = RPNHead(
in_channels=backbone.out_channels,
num_anchors=anchor_generator.num_anchors_per_location()[0],
)
pre_nms_top_n = dict(training=cfg.MODEL.RPN.PRE_NMS_TOPN_TRAIN,
testing=cfg.MODEL.RPN.PRE_NMS_TOPN_TEST)
post_nms_top_n = dict(training=cfg.MODEL.RPN.POST_NMS_TOPN_TRAIN,
testing=cfg.MODEL.RPN.POST_NMS_TOPN_TEST)
rpn = RegionProposalNetwork(
anchor_generator=anchor_generator,
head=head,
fg_iou_thresh=cfg.MODEL.RPN.POS_THRESH_TRAIN,
bg_iou_thresh=cfg.MODEL.RPN.NEG_THRESH_TRAIN,
batch_size_per_image=cfg.MODEL.RPN.BATCH_SIZE_TRAIN,
positive_fraction=cfg.MODEL.RPN.POS_FRAC_TRAIN,
pre_nms_top_n=pre_nms_top_n,
post_nms_top_n=post_nms_top_n,
nms_thresh=cfg.MODEL.RPN.NMS_THRESH,
)
faster_rcnn_predictor = FastRCNNPredictor(2048, 2)
reid_head = deepcopy(box_head)
box_roi_pool = MultiScaleRoIAlign(featmap_names=["feat_res4"],
output_size=14,
sampling_ratio=2)
box_predictor = BBoxRegressor(2048,
num_classes=2,
bn_neck=cfg.MODEL.ROI_HEAD.BN_NECK)
roi_heads = SeqRoIHeads(
# OIM
num_pids=cfg.MODEL.LOSS.LUT_SIZE,
num_cq_size=cfg.MODEL.LOSS.CQ_SIZE,
oim_momentum=cfg.MODEL.LOSS.OIM_MOMENTUM,
oim_scalar=cfg.MODEL.LOSS.OIM_SCALAR,
# SeqNet
faster_rcnn_predictor=faster_rcnn_predictor,
reid_head=reid_head,
# parent class
box_roi_pool=box_roi_pool,
box_head=box_head,
box_predictor=box_predictor,
fg_iou_thresh=cfg.MODEL.ROI_HEAD.POS_THRESH_TRAIN,
bg_iou_thresh=cfg.MODEL.ROI_HEAD.NEG_THRESH_TRAIN,
batch_size_per_image=cfg.MODEL.ROI_HEAD.BATCH_SIZE_TRAIN,
positive_fraction=cfg.MODEL.ROI_HEAD.POS_FRAC_TRAIN,
bbox_reg_weights=None,
score_thresh=cfg.MODEL.ROI_HEAD.SCORE_THRESH_TEST,
nms_thresh=cfg.MODEL.ROI_HEAD.NMS_THRESH_TEST,
detections_per_img=cfg.MODEL.ROI_HEAD.DETECTIONS_PER_IMAGE_TEST,
)
transform = GeneralizedRCNNTransform(
min_size=cfg.INPUT.MIN_SIZE,
max_size=cfg.INPUT.MAX_SIZE,
image_mean=[0.485, 0.456, 0.406],
image_std=[0.229, 0.224, 0.225],
)
self.backbone = backbone
self.rpn = rpn
self.roi_heads = roi_heads
self.transform = transform
# loss weights
self.lw_rpn_reg = cfg.SOLVER.LW_RPN_REG
self.lw_rpn_cls = cfg.SOLVER.LW_RPN_CLS
self.lw_proposal_reg = cfg.SOLVER.LW_PROPOSAL_REG
self.lw_proposal_cls = cfg.SOLVER.LW_PROPOSAL_CLS
self.lw_box_reg = cfg.SOLVER.LW_BOX_REG
self.lw_box_cls = cfg.SOLVER.LW_BOX_CLS
self.lw_box_reid = cfg.SOLVER.LW_BOX_REID
def __init__(
self,
num_classes=2,
# transform parameters
backbone_name='resnet50',
min_size=256,
max_size=512,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
rpn_score_thresh=0.0,
# Box parameters
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.5,
box_detections_per_img=100,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.5,
box_batch_size_per_image=512,
box_positive_fraction=0.25,
bbox_reg_weights=None,
# Ellipse regressor
ellipse_roi_pool=None,
ellipse_head=None,
ellipse_predictor=None,
ellipse_loss_metric="gaussian-angle"):
backbone = resnet_fpn_backbone(backbone_name,
pretrained=True,
trainable_layers=5)
# Input image is grayscale -> in_channels = 1 instead of 3 (COCO)
backbone.body.conv1 = Conv2d(1,
64,
kernel_size=(7, 7),
stride=(2, 2),
padding=(3, 3),
bias=False)
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
anchor_sizes = ((32, ), (64, ), (128, ), (256, ), (512, ))
aspect_ratios = ((0.5, 1.0, 2.0), ) * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator,
rpn_head,
rpn_fg_iou_thresh,
rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction,
rpn_pre_nms_top_n,
rpn_post_nms_top_n,
rpn_nms_thresh,
score_thresh=rpn_score_thresh)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(
featmap_names=['0', '1', '2', '3'],
output_size=7,
sampling_ratio=2)
if box_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 1024
box_head = TwoMLPHead(out_channels * resolution**2,
representation_size)
if box_predictor is None:
representation_size = 1024
box_predictor = FastRCNNPredictor(representation_size, num_classes)
if ellipse_roi_pool is None:
ellipse_roi_pool = MultiScaleRoIAlign(
featmap_names=['0', '1', '2', '3'],
output_size=7,
sampling_ratio=2)
if ellipse_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 1024
ellipse_head = TwoMLPHead(out_channels * resolution**2,
representation_size)
if ellipse_predictor is None:
representation_size = 1024
ellipse_predictor = EllipseRegressor(representation_size,
num_classes)
roi_heads = EllipseRoIHeads(
# Box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img,
# Ellipse
ellipse_roi_pool=ellipse_roi_pool,
ellipse_head=ellipse_head,
ellipse_predictor=ellipse_predictor,
ellipse_loss_metric=ellipse_loss_metric)
if image_mean is None:
image_mean = [0.156]
if image_std is None:
image_std = [0.272]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super().__init__(backbone, rpn, roi_heads, transform)
class RetinaNet(nn.Module):
"""
Implements RetinaNet.
The input to the model is expected to be a list of tensors, each of shape [C, H, W], one for each
image, and should be in 0-1 range. Different images can have different sizes.
The behavior of the model changes depending if it is in training or evaluation mode.
During training, the model expects both the input tensors, as well as a targets (list of dictionary),
containing:
- boxes (``FloatTensor[N, 4]``): the ground-truth boxes in ``[x1, y1, x2, y2]`` format, with
``0 <= x1 < x2 <= W`` and ``0 <= y1 < y2 <= H``.
- labels (Int64Tensor[N]): the class label for each ground-truth box
The model returns a Dict[Tensor] during training, containing the classification and regression
losses.
During inference, the model requires only the input tensors, and returns the post-processed
predictions as a List[Dict[Tensor]], one for each input image. The fields of the Dict are as
follows:
- boxes (``FloatTensor[N, 4]``): the predicted boxes in ``[x1, y1, x2, y2]`` format, with
``0 <= x1 < x2 <= W`` and ``0 <= y1 < y2 <= H``.
- labels (Int64Tensor[N]): the predicted labels for each image
- scores (Tensor[N]): the scores for each prediction
Args:
backbone (nn.Module): the network used to compute the features for the model.
It should contain an out_channels attribute, which indicates the number of output
channels that each feature map has (and it should be the same for all feature maps).
The backbone should return a single Tensor or an OrderedDict[Tensor].
num_classes (int): number of output classes of the model (including the background).
min_size (int): minimum size of the image to be rescaled before feeding it to the backbone
max_size (int): maximum size of the image to be rescaled before feeding it to the backbone
image_mean (Tuple[float, float, float]): mean values used for input normalization.
They are generally the mean values of the dataset on which the backbone has been trained
on
image_std (Tuple[float, float, float]): std values used for input normalization.
They are generally the std values of the dataset on which the backbone has been trained on
anchor_generator (AnchorGenerator): module that generates the anchors for a set of feature
maps.
head (nn.Module): Module run on top of the feature pyramid.
Defaults to a module containing a classification and regression module.
score_thresh (float): Score threshold used for postprocessing the detections.
nms_thresh (float): NMS threshold used for postprocessing the detections.
detections_per_img (int): Number of best detections to keep after NMS.
fg_iou_thresh (float): minimum IoU between the anchor and the GT box so that they can be
considered as positive during training.
bg_iou_thresh (float): maximum IoU between the anchor and the GT box so that they can be
considered as negative during training.
topk_candidates (int): Number of best detections to keep before NMS.
Example:
>>> import torch
>>> import torchvision
>>> from torchvision.models.detection import RetinaNet
>>> from torchvision.models.detection.anchor_utils import AnchorGenerator
>>> # load a pre-trained model for classification and return
>>> # only the features
>>> backbone = torchvision.models.mobilenet_v2(pretrained=True).features
>>> # RetinaNet needs to know the number of
>>> # output channels in a backbone. For mobilenet_v2, it's 1280
>>> # so we need to add it here
>>> backbone.out_channels = 1280
>>>
>>> # let's make the network generate 5 x 3 anchors per spatial
>>> # location, with 5 different sizes and 3 different aspect
>>> # ratios. We have a Tuple[Tuple[int]] because each feature
>>> # map could potentially have different sizes and
>>> # aspect ratios
>>> anchor_generator = AnchorGenerator(
>>> sizes=((32, 64, 128, 256, 512),),
>>> aspect_ratios=((0.5, 1.0, 2.0),)
>>> )
>>>
>>> # put the pieces together inside a RetinaNet model
>>> model = RetinaNet(backbone,
>>> num_classes=2,
>>> anchor_generator=anchor_generator)
>>> model.eval()
>>> x = [torch.rand(3, 300, 400), torch.rand(3, 500, 400)]
>>> predictions = model(x)
"""
__annotations__ = {
'box_coder': det_utils.BoxCoder,
'proposal_matcher': det_utils.Matcher,
}
def __init__(
self,
backbone,
num_classes,
# transform parameters
min_size=800,
max_size=1333,
image_mean=None,
image_std=None,
# Anchor parameters
anchor_generator=None,
head=None,
proposal_matcher=None,
score_thresh=0.05,
nms_thresh=0.5,
detections_per_img=300,
fg_iou_thresh=0.5,
bg_iou_thresh=0.4,
topk_candidates=1000):
super().__init__()
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
self.backbone = backbone
assert isinstance(anchor_generator, (AnchorGenerator, type(None)))
if anchor_generator is None:
anchor_sizes = tuple(
(x, int(x * 2**(1.0 / 3)), int(x * 2**(2.0 / 3)))
for x in [32, 64, 128, 256, 512])
aspect_ratios = ((0.5, 1.0, 2.0), ) * len(anchor_sizes)
anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
self.anchor_generator = anchor_generator
if head is None:
head = RetinaNetHead(
backbone.out_channels,
anchor_generator.num_anchors_per_location()[0], num_classes)
self.head = head
if proposal_matcher is None:
proposal_matcher = det_utils.Matcher(
fg_iou_thresh,
bg_iou_thresh,
allow_low_quality_matches=True,
)
self.proposal_matcher = proposal_matcher
self.box_coder = det_utils.BoxCoder(weights=(1.0, 1.0, 1.0, 1.0))
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
self.transform = GeneralizedRCNNTransform(min_size, max_size,
image_mean, image_std)
self.score_thresh = score_thresh
self.nms_thresh = nms_thresh
self.detections_per_img = detections_per_img
self.topk_candidates = topk_candidates
# used only on torchscript mode
self._has_warned = False
@torch.jit.unused
def eager_outputs(self, losses, detections):
# type: (Dict[str, Tensor], List[Dict[str, Tensor]]) -> Tuple[Dict[str, Tensor], List[Dict[str, Tensor]]]
if self.training:
return losses
return detections
def compute_loss(self, targets, head_outputs, anchors):
# type: (List[Dict[str, Tensor]], Dict[str, Tensor], List[Tensor]) -> Dict[str, Tensor]
matched_idxs = []
for anchors_per_image, targets_per_image in zip(anchors, targets):
if targets_per_image['boxes'].numel() == 0:
matched_idxs.append(
torch.full((anchors_per_image.size(0), ),
-1,
dtype=torch.int64,
device=anchors_per_image.device))
continue
match_quality_matrix = box_ops.box_iou(targets_per_image['boxes'],
anchors_per_image)
matched_idxs.append(self.proposal_matcher(match_quality_matrix))
return self.head.compute_loss(targets, head_outputs, anchors,
matched_idxs)
def postprocess_detections(self, head_outputs, anchors, image_shapes):
# type: (Dict[str, List[Tensor]], List[List[Tensor]], List[Tuple[int, int]]) -> List[Dict[str, Tensor]]
class_logits = head_outputs['cls_logits']
box_regression = head_outputs['bbox_regression']
num_images = len(image_shapes)
detections: List[Dict[str, Tensor]] = []
for index in range(num_images):
box_regression_per_image = [br[index] for br in box_regression]
logits_per_image = [cl[index] for cl in class_logits]
anchors_per_image, image_shape = anchors[index], image_shapes[
index]
image_boxes = []
image_scores = []
image_labels = []
for box_regression_per_level, logits_per_level, anchors_per_level in \
zip(box_regression_per_image, logits_per_image, anchors_per_image):
num_classes = logits_per_level.shape[-1]
# remove low scoring boxes
scores_per_level = torch.sigmoid(logits_per_level).flatten()
keep_idxs = scores_per_level > self.score_thresh
scores_per_level = scores_per_level[keep_idxs]
topk_idxs = torch.where(keep_idxs)[0]
# keep only topk scoring predictions
num_topk = min(self.topk_candidates, topk_idxs.size(0))
scores_per_level, idxs = scores_per_level.topk(num_topk)
topk_idxs = topk_idxs[idxs]
anchor_idxs = torch.div(topk_idxs,
num_classes,
rounding_mode='floor')
labels_per_level = topk_idxs % num_classes
boxes_per_level = self.box_coder.decode_single(
box_regression_per_level[anchor_idxs],
anchors_per_level[anchor_idxs])
boxes_per_level = box_ops.clip_boxes_to_image(
boxes_per_level, image_shape)
image_boxes.append(boxes_per_level)
image_scores.append(scores_per_level)
image_labels.append(labels_per_level)
image_boxes = torch.cat(image_boxes, dim=0)
image_scores = torch.cat(image_scores, dim=0)
image_labels = torch.cat(image_labels, dim=0)
# non-maximum suppression
keep = box_ops.batched_nms(image_boxes, image_scores, image_labels,
self.nms_thresh)
keep = keep[:self.detections_per_img]
detections.append({
'boxes': image_boxes[keep],
'scores': image_scores[keep],
'labels': image_labels[keep],
})
return detections
def forward(self, images, targets=None):
# type: (List[Tensor], Optional[List[Dict[str, Tensor]]]) -> Tuple[Dict[str, Tensor], List[Dict[str, Tensor]]]
"""
Args:
images (list[Tensor]): images to be processed
targets (list[Dict[Tensor]]): ground-truth boxes present in the image (optional)
Returns:
result (list[BoxList] or dict[Tensor]): the output from the model.
During training, it returns a dict[Tensor] which contains the losses.
During testing, it returns list[BoxList] contains additional fields
like `scores`, `labels` and `mask` (for Mask R-CNN models).
"""
if self.training and targets is None:
raise ValueError("In training mode, targets should be passed")
if self.training:
assert targets is not None
for target in targets:
boxes = target["boxes"]
if isinstance(boxes, torch.Tensor):
if len(boxes.shape) != 2 or boxes.shape[-1] != 4:
raise ValueError("Expected target boxes to be a tensor"
"of shape [N, 4], got {:}.".format(
boxes.shape))
else:
raise ValueError("Expected target boxes to be of type "
"Tensor, got {:}.".format(type(boxes)))
# get the original image sizes
original_image_sizes: List[Tuple[int, int]] = []
for img in images:
val = img.shape[-2:]
assert len(val) == 2
original_image_sizes.append((val[0], val[1]))
# transform the input
images, targets = self.transform(images, targets)
# Check for degenerate boxes
# TODO: Move this to a function
if targets is not None:
for target_idx, target in enumerate(targets):
boxes = target["boxes"]
degenerate_boxes = boxes[:, 2:] <= boxes[:, :2]
if degenerate_boxes.any():
# print the first degenerate box
bb_idx = torch.where(degenerate_boxes.any(dim=1))[0][0]
degen_bb: List[float] = boxes[bb_idx].tolist()
raise ValueError(
"All bounding boxes should have positive height and width."
" Found invalid box {} for target at index {}.".format(
degen_bb, target_idx))
# get the features from the backbone
features = self.backbone(images.tensors)
if isinstance(features, torch.Tensor):
features = OrderedDict([('0', features)])
# TODO: Do we want a list or a dict?
features = list(features.values())
for idx in range(len(features)):
features[idx] = features[idx].to(torch.float32)
# compute the retinanet heads outputs using the features
head_outputs = self.head(features)
for key in head_outputs:
head_outputs[key] = head_outputs[key].to(torch.float32)
# create the set of anchors
anchors = self.anchor_generator(images, features)
losses = {}
detections: List[Dict[str, Tensor]] = []
if self.training:
assert targets is not None
# compute the losses
losses = self.compute_loss(targets, head_outputs, anchors)
else:
# recover level sizes
num_anchors_per_level = [x.size(2) * x.size(3) for x in features]
HW = 0
for v in num_anchors_per_level:
HW += v
HWA = head_outputs['cls_logits'].size(1)
A = HWA // HW
num_anchors_per_level = [hw * A for hw in num_anchors_per_level]
# split outputs per level
split_head_outputs: Dict[str, List[Tensor]] = {}
for k in head_outputs:
split_head_outputs[k] = list(head_outputs[k].split(
num_anchors_per_level, dim=1))
split_anchors = [
list(a.split(num_anchors_per_level)) for a in anchors
]
# compute the detections
detections = self.postprocess_detections(split_head_outputs,
split_anchors,
images.image_sizes)
detections = self.transform.postprocess(detections,
images.image_sizes,
original_image_sizes)
if torch.jit.is_scripting():
if not self._has_warned:
warnings.warn(
"RetinaNet always returns a (Losses, Detections) tuple in scripting"
)
self._has_warned = True
return losses, detections
return self.eager_outputs(losses, detections)
from torchvision.datasets.coco import CocoDetection
from torchvision.models.detection.transform import GeneralizedRCNNTransform
from torchvision.transforms import transforms
from pdetection import transform
from torch.utils.data.dataloader import DataLoader
import torch
from pdetection import utils
path_data = '/home/peng/Documents/srch/Object Detection/dataset/coco/val2017'
path_anno = '/home/peng/Documents/srch/Object Detection/dataset/coco/annotations_trainval2017/annotations/instances_val2017.json'
coco_dset = CocoDetection(root=path_data, annFile=path_anno)
trans = transform.ODTransformer(800, 1333, [0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])
trans_compare = GeneralizedRCNNTransform(800, 1333, [0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])
# trans = transform.ODTransformer(800, 1333, [0.4, 0.5, 0.6], [0.5, 1, 2])
cocoloader = DataLoader(coco_dset, batch_size=3, collate_fn=utils.collate_fn)
for data in cocoloader:
images, targets = data
inputimgs = []
totensor = transforms.ToTensor()
for img in images:
inputimgs.append(totensor(img))
src_img_sizes = [img.shape[-2:] for img in inputimgs]
# print([img.shape[-2:] for img in inputimgs])
targets1 = utils.totargets(targets)
targets2 = utils.totargets(targets)
print(targets2[2]['boxes'])
inputimgs_bk = [torch.zeros_like(img).copy_(img) for img in inputimgs]
def __init__(
self,
backbone,
num_classes=None,
num_pids=5532,
num_cq_size=5000,
# transform parameters
min_size=900,
max_size=1500,
image_mean=None,
image_std=None,
# Anchor settings:
anchor_scales=None,
anchor_ratios=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=12000,
rpn_pre_nms_top_n_test=6000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=300,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
# Box parameters
box_roi_pool=None,
feat_head=None,
box_predictor=None,
box_score_thresh=0.0,
box_nms_thresh=0.4,
box_detections_per_img=300,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.1,
box_batch_size_per_image=128,
box_positive_fraction=0.5,
bbox_reg_weights=None,
# ReID parameters
embedding_head=None,
reid_loss=None):
if not hasattr(backbone, "out_channels"):
raise ValueError(
'backbone should contain an attribute out_channels '
'specifying the number of output channels (assumed to be the '
'same for all the levels)')
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
'num_classes should be None when box_predictor is specified'
)
else:
if box_predictor is None:
raise ValueError(
'num_classes should not be None when box_predictor'
'is not specified')
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
if anchor_scales is None:
anchor_scales = ((32, 64, 128, 256, 512), )
if anchor_ratios is None:
anchor_ratios = ((0.5, 1.0, 2.0), )
rpn_anchor_generator = AnchorGenerator(anchor_scales,
anchor_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = self._set_rpn(rpn_anchor_generator, rpn_head, rpn_fg_iou_thresh,
rpn_bg_iou_thresh, rpn_batch_size_per_image,
rpn_positive_fraction, rpn_pre_nms_top_n,
rpn_post_nms_top_n, rpn_nms_thresh)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(featmap_names=['feat_res4'],
output_size=14,
sampling_ratio=2)
if feat_head is None:
raise ValueError('feat_head should be specified manually.')
# resolution = box_roi_pool.output_size[0]
# representation_size = 2048
# # ConvHead should be part of the backbone
# # feat_head = TwoMLPHead(
# # out_channels * resolution ** 2,
# # representation_size)
if box_predictor is None:
box_predictor = CoordRegressor(2048, num_classes)
if embedding_head is None:
embedding_head = ReIDEmbeddingProj(
featmap_names=['feat_res4', 'feat_res5'],
in_channels=[1024, 2048],
dim=256)
if reid_loss is None:
reid_loss = HOIMLoss(256, num_pids, num_cq_size, 0.5, 30.0)
roi_heads = self._set_roi_heads(
embedding_head, reid_loss, box_roi_pool, feat_head, box_predictor,
box_fg_iou_thresh, box_bg_iou_thresh, box_batch_size_per_image,
box_positive_fraction, bbox_reg_weights, box_score_thresh,
box_nms_thresh, box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(FasterRCNN_HOIM, self).__init__(backbone, rpn, roi_heads,
transform)
def __init__(
self,
backbone,
n_channel_backbone=5,
num_classes=None,
# transform parameters
min_size=800,
max_size=1333,
#min_size=720, max_size=1280,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.5,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
anchor_sizes=[32, 64, 128, 256, 512],
# Box parameters
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.4,
box_detections_per_img=30,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.5,
box_batch_size_per_image=512,
box_positive_fraction=0.25,
bbox_reg_weights=None,
weight_loss=False,
use_soft_nms=False,
use_context=False,
use_track_branch=False):
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
ratios = ((0.5, 1.0, 2.0), )
aspect_ratios = ratios * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator, rpn_head,
rpn_fg_iou_thresh, rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction, rpn_pre_nms_top_n,
rpn_post_nms_top_n, rpn_nms_thresh,
weight_loss)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(featmap_names=[0, 1, 2, 3],
output_size=7,
sampling_ratio=2)
if n_channel_backbone == 6:
box_roi_pool = MultiScaleRoIAlign(
featmap_names=[0, 1, 2, 3, 4],
output_size=7,
sampling_ratio=2)
representation_size1 = 1024
representation_size2 = 1024
track_embedding_size = 1024
if box_head is None:
resolution = box_roi_pool.output_size[0]
if use_context:
box_head = TwoMLPHead(2 * out_channels * resolution**2,
representation_size1,
representation_size2)
else:
box_head = TwoMLPHead(out_channels * resolution**2,
representation_size1,
representation_size2)
if use_track_branch:
if use_context:
track_embedding = TwoMLPHead(2 * out_channels * resolution**2,
representation_size1,
track_embedding_size)
else:
track_embedding = TwoMLPHead(out_channels * resolution**2,
representation_size1,
track_embedding_size)
else:
track_embedding = None
if box_predictor is None:
box_predictor = FastRCNNPredictor(representation_size1,
num_classes)
if num_classes > 2:
use_soft_nms = False
roi_heads = RoIHeads(
# Box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img,
weight_loss=weight_loss,
use_soft_nms=use_soft_nms,
use_context=use_context)
if use_track_branch:
track_heads = TrackHeads(box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
weight_loss=False,
use_context=False,
track_embedding=track_embedding)
else:
track_heads = None
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(FasterRCNN, self).__init__(backbone, rpn, roi_heads, track_heads,
transform, n_channel_backbone)
def __init__(
self,
backbone: nn.Module,
num_classes: int,
anchor_grids: List[List[int]],
# transform parameters
min_size: int = 320,
max_size: int = 416,
image_mean: Optional[List[float]] = None,
image_std: Optional[List[float]] = None,
# Anchor parameters
anchor_generator: Optional[nn.Module] = None,
head: Optional[nn.Module] = None,
# Training parameter
compute_loss: Optional[nn.Module] = None,
fg_iou_thresh: float = 0.5,
bg_iou_thresh: float = 0.4,
# Post Process parameter
postprocess_detections: Optional[nn.Module] = None,
score_thresh: float = 0.05,
nms_thresh: float = 0.5,
detections_per_img: int = 300,
):
super().__init__()
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
self.backbone = backbone
if anchor_generator is None:
strides: List[int] = [8, 16, 32]
anchor_generator = AnchorGenerator(strides, anchor_grids)
self.anchor_generator = anchor_generator
if compute_loss is None:
compute_loss = SetCriterion(
weights=(1.0, 1.0, 1.0, 1.0),
fg_iou_thresh=fg_iou_thresh,
bg_iou_thresh=bg_iou_thresh,
)
self.compute_loss = compute_loss
if head is None:
head = YoloHead(
backbone.out_channels,
anchor_generator.num_anchors,
num_classes,
)
self.head = head
if image_mean is None:
image_mean = [0., 0., 0.]
if image_std is None:
image_std = [1., 1., 1.]
self.transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
if postprocess_detections is None:
postprocess_detections = PostProcess(score_thresh, nms_thresh, detections_per_img)
self.postprocess_detections = postprocess_detections
# used only on torchscript mode
self._has_warned = False
default=8,
help='Index to the dataset for an example')
parser.add_argument('--outdir',
type=str,
default='examples_detection',
help='Folder for output images')
if __name__ == '__main__':
args = parser.parse_args()
device = 'cuda' if torch.cuda.is_available() else 'cpu'
root = args.root
annfile = args.annfile
# Load a maskRCNN finetuned on our birds
network_transform = GeneralizedRCNNTransform(800, 1333, (0, 0, 0),
(1, 1, 1))
backbone = resnet_fpn_backbone(backbone_name='resnet101', pretrained=False)
model = MaskRCNN(backbone, num_classes=2)
model.transform = network_transform
model.eval()
model.load_state_dict(torch.load('models/detector.pth'))
model.to(device)
# Load a data split
normalize = T.Normalize(mean=[102.9801, 115.9465, 122.7717],
std=[1., 1., 1.])
coco = COCO(annfile)
# Load an image example
available_Ids = coco.getImgIds()
imgfile = coco.loadImgs(available_Ids[args.index])[0]['file_name']
def __init__(
self,
backbone,
num_classes=None,
# transform parameters
min_size=800,
max_size=1333,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
# Box parameters
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.5,
box_detections_per_img=100,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.5,
box_batch_size_per_image=512,
box_positive_fraction=0.25,
bbox_reg_weights=None):
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
anchor_sizes = ((32, ), (64, ), (128, ), (256, ), (512, ))
aspect_ratios = ((0.5, 1.0, 2.0), ) * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator, rpn_head,
rpn_fg_iou_thresh, rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction, rpn_pre_nms_top_n,
rpn_post_nms_top_n, rpn_nms_thresh)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(
featmap_names=['0', '1', '2', '3'],
output_size=7,
sampling_ratio=2)
if box_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 1024
box_head = TwoMLPHead(out_channels * resolution**2,
representation_size)
if box_predictor is None:
representation_size = 1024
box_predictor = FastRCNNPredictor(representation_size, num_classes)
roi_heads = RoIHeads(
# Box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(FasterRCNN, self).__init__(backbone, rpn, roi_heads, transform)
import torch
import torchvision
from torch.jit.annotations import Tuple, List, Dict, Optional
import numpy as np
import cv2
import dataset
images, targets = dataset.load_data()
to_tensor = torchvision.transforms.ToTensor()
images = [to_tensor(image) for image in images]
targets = [{
'boxes': item['boxes'],
'labels': item['labels']
} for item in targets]
min_size = [800, 820, 900]
max_size = 1333
from torchvision.models.detection.transform import GeneralizedRCNNTransform
## transform
image_mean = [0.485, 0.456, 0.406]
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
transform.train()
print(transform.training)
images, targets = transform(images, targets)
print(images.tensors.shape)
print(images.image_sizes)
print(len(targets))
print(targets[0])
def __init__(
self,
backbone,
num_classes=None,
# transform parameters
min_size=800,
max_size=1333,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
# Box parameters
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.3,
box_detections_per_img=128,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.5,
box_batch_size_per_image=64,
box_positive_fraction=0.25,
bbox_reg_weights=None):
print("Using modified Faster RCNN....")
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if box_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 1024
box_head = TwoMLPHead(out_channels * resolution**2,
representation_size)
if box_predictor is None:
representation_size = 1024
box_predictor = FastRCNNPredictor(representation_size, num_classes)
rpn = None
roi_heads = RoIHeads(
# Box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super().__init__(backbone, rpn, roi_heads, transform)
import pickle
import random
import argparse
import numpy as np
from datetime import datetime
import torch
import torch.nn as nn
import torch.optim as optim
from models import get_model
from datasets import get_loader
from helper import compute_ts_road_map, compute_ats_bounding_boxes
from torchvision.models.detection.transform import GeneralizedRCNNTransform
target_transform = GeneralizedRCNNTransform(800, 800, [0., 0., 0.],
[1., 1., 1.])
def get_mask_ts(mask, target):
mask = nn.Sigmoid()(mask) > 0.5
temp_tensor = torch.zeros(1, 3, 400, 400)
temp_target = [{'masks': mask, 'boxes': torch.tensor([[1., 1., 1., 1.]])}]
_, temp_target = target_transform(temp_tensor, temp_target)
predicted_road_map = temp_target[0]['masks'][0, :1]
ts_road_map = compute_ts_road_map(predicted_road_map, target[0]['masks'])
return ts_road_map
def get_detection_ts(detection, target):
class YOLO(nn.Module):
def __init__(
self,
backbone: nn.Module,
num_classes: int,
anchor_grids: List[List[int]],
# transform parameters
min_size: int = 320,
max_size: int = 416,
image_mean: Optional[List[float]] = None,
image_std: Optional[List[float]] = None,
# Anchor parameters
anchor_generator: Optional[nn.Module] = None,
head: Optional[nn.Module] = None,
# Training parameter
compute_loss: Optional[nn.Module] = None,
fg_iou_thresh: float = 0.5,
bg_iou_thresh: float = 0.4,
# Post Process parameter
postprocess_detections: Optional[nn.Module] = None,
score_thresh: float = 0.05,
nms_thresh: float = 0.5,
detections_per_img: int = 300,
):
super().__init__()
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
self.backbone = backbone
if anchor_generator is None:
strides: List[int] = [8, 16, 32]
anchor_generator = AnchorGenerator(strides, anchor_grids)
self.anchor_generator = anchor_generator
if compute_loss is None:
compute_loss = SetCriterion(
weights=(1.0, 1.0, 1.0, 1.0),
fg_iou_thresh=fg_iou_thresh,
bg_iou_thresh=bg_iou_thresh,
)
self.compute_loss = compute_loss
if head is None:
head = YoloHead(
backbone.out_channels,
anchor_generator.num_anchors,
num_classes,
)
self.head = head
if image_mean is None:
image_mean = [0., 0., 0.]
if image_std is None:
image_std = [1., 1., 1.]
self.transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
if postprocess_detections is None:
postprocess_detections = PostProcess(score_thresh, nms_thresh, detections_per_img)
self.postprocess_detections = postprocess_detections
# used only on torchscript mode
self._has_warned = False
@torch.jit.unused
def eager_outputs(
self,
losses: Dict[str, Tensor],
detections: List[Dict[str, Tensor]],
) -> Tuple[Dict[str, Tensor], List[Dict[str, Tensor]]]:
if self.training:
return losses
return detections
def forward(
self,
images: List[Tensor],
targets: Optional[List[Dict[str, Tensor]]] = None,
) -> Tuple[Dict[str, Tensor], List[Dict[str, Tensor]]]:
"""
Arguments:
images (list[Tensor]): images to be processed
targets (list[Dict[Tensor]]): ground-truth boxes present in the image (optional)
Returns:
result (list[BoxList] or dict[Tensor]): the output from the model.
During Training, it returns a dict[Tensor] which contains the losses
TODO, currently this repo doesn't support training.
During Testing, it returns list[BoxList] contains additional fields
like `scores` and `labels`.
"""
# get the original image sizes
original_image_sizes = torch.jit.annotate(List[Tuple[int, int]], [])
for img in images:
val = img.shape[-2:]
assert len(val) == 2
original_image_sizes.append((val[0], val[1]))
# transform the input
images, targets = self.transform(images, targets)
# get the features from the backbone
features = self.backbone(images.tensors)
# compute the yolo heads outputs using the features
head_outputs = self.head(features)
# create the set of anchors
anchors_tuple = self.anchor_generator(features)
losses = {}
detections = torch.jit.annotate(List[Dict[str, Tensor]], [])
if self.training:
assert targets is not None
# compute the losses
losses = self.compute_loss(targets, head_outputs, anchors_tuple[0])
else:
# compute the detections
detections = self.postprocess_detections(head_outputs, anchors_tuple, images.image_sizes)
detections = self.transform.postprocess(detections, images.image_sizes, original_image_sizes)
if torch.jit.is_scripting():
if not self._has_warned:
warnings.warn("YOLO always returns a (Losses, Detections) tuple in scripting")
self._has_warned = True
return losses, detections
else:
return self.eager_outputs(losses, detections)