def test_transforms_presets_ssd():
im_fname = gcv.utils.download('https://github.com/dmlc/web-data/blob/master/' +
'gluoncv/detection/biking.jpg?raw=true', path='biking.jpg')
x, orig_img = ssd.load_test(im_fname, short=512)
x1, orig_img1 = ssd.transform_test(mx.image.imread(im_fname), short=512)
np.testing.assert_allclose(x.asnumpy(), x1.asnumpy())
np.testing.assert_allclose(orig_img, orig_img1)
if not osp.isdir(osp.expanduser('~/.mxnet/datasets/voc')):
return
train_dataset = gcv.data.VOCDetection(splits=((2007, 'trainval'), (2012, 'trainval')))
val_dataset = gcv.data.VOCDetection(splits=[(2007, 'test')])
width, height = (512, 512)
net = gcv.model_zoo.get_model('ssd_512_resnet50_v1_voc', pretrained=False, pretrained_base=False)
net.initialize()
num_workers = 0
batch_size = 4
with autograd.train_mode():
_, _, anchors = net(mx.nd.zeros((1, 3, height, width)))
batchify_fn = Tuple(Stack(), Stack(), Stack()) # stack image, cls_targets, box_targets
train_loader = gluon.data.DataLoader(
train_dataset.transform(ssd.SSDDefaultTrainTransform(width, height, anchors)),
batch_size, True, batchify_fn=batchify_fn, last_batch='rollover', num_workers=num_workers)
val_batchify_fn = Tuple(Stack(), Pad(pad_val=-1))
val_loader = gluon.data.DataLoader(
val_dataset.transform(ssd.SSDDefaultValTransform(width, height)),
batch_size, False, batchify_fn=val_batchify_fn, last_batch='keep', num_workers=num_workers)
train_loader2 = gluon.data.DataLoader(
train_dataset.transform(ssd.SSDDefaultTrainTransform(width, height)),
batch_size, True, batchify_fn=val_batchify_fn, last_batch='rollover', num_workers=num_workers)
for loader in [train_loader, val_loader, train_loader2]:
for i, batch in enumerate(loader):
if i > 1:
break
pass
def test_transforms_presets_ssd():
im_fname = gcv.utils.download('https://github.com/dmlc/web-data/blob/master/' +
'gluoncv/detection/biking.jpg?raw=true', path='biking.jpg')
x, orig_img = ssd.load_test(im_fname, short=512)
x1, orig_img1 = ssd.transform_test(mx.image.imread(im_fname), short=512)
np.testing.assert_allclose(x.asnumpy(), x1.asnumpy())
np.testing.assert_allclose(orig_img, orig_img1)
if not osp.isdir(osp.expanduser('~/.mxnet/datasets/voc')):
return
train_dataset = VOCDetectionTiny()
val_dataset = VOCDetectionTiny(splits=[('tiny_motorbike', 'test')])
width, height = (512, 512)
net = gcv.model_zoo.get_model('ssd_512_resnet50_v1_voc', pretrained=False, pretrained_base=False)
net.initialize()
num_workers = 0
batch_size = 4
with autograd.train_mode():
_, _, anchors = net(mx.nd.zeros((1, 3, height, width)))
batchify_fn = Tuple(Stack(), Stack(), Stack()) # stack image, cls_targets, box_targets
train_loader = gluon.data.DataLoader(
train_dataset.transform(ssd.SSDDefaultTrainTransform(width, height, anchors)),
batch_size, True, batchify_fn=batchify_fn, last_batch='rollover', num_workers=num_workers)
val_batchify_fn = Tuple(Stack(), Pad(pad_val=-1))
val_loader = gluon.data.DataLoader(
val_dataset.transform(ssd.SSDDefaultValTransform(width, height)),
batch_size, False, batchify_fn=val_batchify_fn, last_batch='keep', num_workers=num_workers)
train_loader2 = gluon.data.DataLoader(
train_dataset.transform(ssd.SSDDefaultTrainTransform(width, height)),
batch_size, True, batchify_fn=val_batchify_fn, last_batch='rollover', num_workers=num_workers)
for loader in [train_loader, val_loader, train_loader2]:
for i, batch in enumerate(loader):
if i > 1:
break
pass
def get_full_frame_info(a_frame):
x, frame = transform_test(a_frame, short=512)
class_IDs, scores, bounding_boxs = detector(x)
pose_input, upscale_bbox = detector_to_simple_pose(frame, class_IDs, scores, bounding_boxs)
if len(upscale_bbox)>0:
predicted_heatmap = pose_net(pose_input)
pred_coords, confidence = heatmap_to_coord(predicted_heatmap, upscale_bbox)
pred_coords = pred_coords.asnumpy()
return class_IDs, scores, upscale_bbox, pred_coords, confidence, bounding_boxs
def detection(net, image, use_gpu):
'''
进行预测：
input:
net(dict): 模型
image(str): 图片(numpy)
use_gpu(bool): 是否使用gpu
return:
pred(dict): 包含各种信息的字典（若未检测到人则该返回值为None）
img(numpy): 图片
'''
if use_gpu:
ctx = mx.gpu()
else:
ctx = mx.cpu()
img_adarry = nd.array(image)
x, img = transform_test(img_adarry,
short=512,
max_size=1024,
mean=(0.485, 0.456, 0.406),
std=(0.229, 0.224, 0.225))
if use_gpu:
# 转移至GPU
x = x.as_in_context(ctx)
class_IDs, scores, bounding_boxs = net['detector'](x)
pose_input, upscale_bbox = detector_to_simple_pose(img, class_IDs, scores,
bounding_boxs)
if len(upscale_bbox) == 0:
# 图片中未检测到人
return None, img
if use_gpu:
# 转移至GPU
pose_input = pose_input.as_in_context(ctx)
predicted_heatmap = net['pose_net'](pose_input)
pred_coords, confidence = heatmap_to_coord(predicted_heatmap, upscale_bbox)
if use_gpu:
# 转移至GPU
pred_coords = pred_coords.as_in_context(ctx)
pred = {
'class_IDs': class_IDs,
'scores': scores,
'bounding_boxs': bounding_boxs,
'pred_coords': pred_coords,
'confidence': confidence
}
return pred, img
def preprocessing(input_img, **kwargs):
to_normalize = False if np.percentile(input_img, 98) > 1.0 else True
if len(input_img.shape) == 4:
print(
"Only preprocessing single image, we will consider the first one of the batch"
)
image = input_img[0] * 255.0 if to_normalize else input_img[0] * 1.0
else:
image = input_img * 255.0 if to_normalize else input_img * 1.0
image = cv2.resize(image, input_shape)
x, _ = transform_test(mx.nd.array(image), min(input_shape))
return x
def get_skeleton_from_frame(a_frame):
ok_flag = False
extra_person_flag = False
x, frame = transform_test(a_frame, short=512)
class_IDs, scores, bounding_boxs = detector(x)
pose_input, upscale_bbox = detector_to_simple_pose(frame, class_IDs, scores, bounding_boxs)
b_coords = 0
if len(upscale_bbox)>0:
predicted_heatmap = pose_net(pose_input)
pred_coords, confidence = heatmap_to_coord(predicted_heatmap, upscale_bbox)
pred_coords = pred_coords.asnumpy()
b_coords = pred_coords[0]
if pred_coords.shape[0]>2:
extra_person_flag = True
if pred_coords.shape[0]==2:
# doing best guest when two boxes ( subject and background are similar)
if upscale_bbox[0][3]==512:
b_coords = pred_coords[1]
ok_flag= True
return ok_flag, extra_person_flag, b_coords
cap1 = cv.VideoCapture(args.input)
cap2 = cv.VideoCapture(args.demo)
# 标准特征
angeleCal = AngeleCal(args.data)
ret1, frame1 = cap1.read()
ret2, frame2 = cap2.read()
while ret1 and ret2:
frame1 = mx.nd.array(cv.cvtColor(frame1, cv.COLOR_BGR2RGB)).astype('uint8')
frame2 = cv.cvtColor(frame2, cv.COLOR_BGR2RGB)
# 目标检测
x, img = transform_test(frame1, short=512, max_size=680)
x = x.as_in_context(ctx)
class_IDs, scores, bounding_boxs = detector(x)
pose_input, upscale_bbox = detector_to_simple_pose(img,
class_IDs,
scores,
bounding_boxs,
ctx=ctx)
# 姿态识别
results = None
if len(upscale_bbox) > 0:
predicted_heatmap = estimator(pose_input)
pred_coords, confidence = heatmap_to_coord(predicted_heatmap,
upscale_bbox)
parser = argparse.ArgumentParser()
parser.add_argument('--input')
parser.add_argument('--output', required=True)
args = parser.parse_args()
# 视频读取
cap = cv.VideoCapture(args.input)
ret, frame = cap.read()
features = []
while ret:
frame = mx.nd.array(cv.cvtColor(frame, cv.COLOR_BGR2RGB)).astype('uint8')
# 目标检测
x, img = transform_test(frame, short=512)
x = x.as_in_context(ctx)
class_IDs, scores, bounding_boxs = detector(x)
pose_input, upscale_bbox = detector_to_simple_pose(img,
class_IDs,
scores,
bounding_boxs,
ctx=ctx)
if len(upscale_bbox) > 0:
predicted_heatmap = estimator(pose_input)
pred_coords, confidence = heatmap_to_coord(predicted_heatmap,
upscale_bbox)
img = cv_plot_keypoints(img, pred_coords, confidence, class_IDs,
bounding_boxs, scores)
