def test_rcnn(imageset,
year,
root_path,
devkit_path,
prefix,
epoch,
ctx,
vis=False,
has_rpn=True,
proposal='rpn',
end2end=False):
# load symbol and testing data
if has_rpn:
sym = get_vgg_test()
config.TEST.HAS_RPN = True
config.TEST.RPN_PRE_NMS_TOP_N = 6000
config.TEST.RPN_POST_NMS_TOP_N = 300
voc, roidb = load_gt_roidb(imageset, year, root_path, devkit_path)
else:
sym = get_vgg_rcnn_test()
voc, roidb = eval('load_test_' + proposal + '_roidb')(imageset, year,
root_path,
devkit_path)
# get test data iter
test_data = ROIIter(roidb, batch_size=1, shuffle=False, mode='test')
# load model
args, auxs, _ = load_param(prefix, epoch, convert=True, ctx=ctx)
# detect
detector = Detector(sym, ctx, args, auxs)
pred_eval(detector, test_data, voc, vis=vis)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
demo_net(predictor, args.image, args.vis)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES, num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
assert os.path.exists(args.image), args.image + ' not found'
im = cv2.imread(args.image)
demo_net(predictor, im, args.vis)
def main():
color = cv2.imread(args.img) # read image in b,g,r order
img, scale = resize(color.copy(), 640, 1024)
im_info = np.array([[img.shape[0], img.shape[1], scale]],
dtype=np.float32) # (h, w, scale)
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 1, 2) # change to r,g,b order
img = img[np.newaxis, :] # extend to (n, c, h, w)
ctx = mx.gpu(args.gpu)
_, arg_params, aux_params = mx.model.load_checkpoint(
args.prefix, args.epoch)
arg_params, aux_params = ch_dev(arg_params, aux_params, ctx)
if 'resnet' in args.prefix:
sym = resnet_50(num_class=2,
bn_mom=0.99,
bn_global=True,
is_train=False)
else:
sym = get_vgg_test(num_classes=2)
arg_params["data"] = mx.nd.array(img, ctx)
arg_params["im_info"] = mx.nd.array(im_info, ctx)
exe = sym.bind(ctx,
arg_params,
args_grad=None,
grad_req="null",
aux_states=aux_params)
exe.forward(is_train=False)
output_dict = {
name: nd
for name, nd in zip(sym.list_outputs(), exe.outputs)
}
rois = output_dict['rpn_rois_output'].asnumpy(
)[:, 1:] # first column is index
scores = output_dict['cls_prob_reshape_output'].asnumpy()[0]
bbox_deltas = output_dict['bbox_pred_reshape_output'].asnumpy()[0]
pred_boxes = bbox_pred(rois, bbox_deltas)
pred_boxes = clip_boxes(pred_boxes, (im_info[0][0], im_info[0][1]))
cls_boxes = pred_boxes[:, 4:8]
cls_scores = scores[:, 1]
keep = np.where(cls_scores >= args.thresh)[0]
cls_boxes = cls_boxes[keep, :]
cls_scores = cls_scores[keep]
dets = np.hstack((cls_boxes, cls_scores[:, np.newaxis])).astype(np.float32)
keep = nms(dets.astype(np.float32), args.nms_thresh)
dets = dets[keep, :]
keep = nest(dets, thresh=args.nest_thresh)
dets = dets[keep, :]
for i in range(dets.shape[0]):
bbox = dets[i, :4]
cv2.rectangle(
color, (int(round(bbox[0] / scale)), int(round(bbox[1] / scale))),
(int(round(bbox[2] / scale)), int(round(bbox[3] / scale))),
(0, 255, 0), 2)
cv2.imwrite("result.jpg", color)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
viz = mx.viz.plot_network(symbol, shape={'data': (1, 3, 600, 800)})
viz.view('faster-rcnn')
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
demo_net(predictor, args.image, args.vis)
def __init__(self,
gpu=0,
prefix=os.path.abspath(
os.path.join(os.path.dirname(__file__), '..', '..',
'data', 'final')),
epoch=0):
ctx = mx.gpu(gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
self.predictor = self.get_net(symbol, prefix, epoch, ctx)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
df_load = pd.read_csv(args.rec)
classes = ['blouse', 'outwear', 'skirt', 'dress', 'trousers']
results = []
for cls in classes:
print('loading the model')
prefix = args.prefix + cls
df_pred = df_load[df_load['image_category'] == cls].copy()
df_pred.reset_index(inplace=True)
del df_pred['index']
predictor = get_net(symbol, prefix, args.epoch, ctx)
record = []
for idx, row in df_pred.iterrows():
image_id = row['image_id']
cat = image_id.split('/')[1]
if cat != cls:
continue
image = os.path.join(data_root, image_id)
#cat = row['image_category']
res = demo_net(predictor, image, cls, args.vis)
print(res)
datum = [
image_id, cls,
int(res[0]),
int(res[1]),
int(res[2]),
int(res[3]), res[4]
]
record.append(datum)
#if idx >10:
# break
if res[4] < 0.1:
print(image_id)
cloumns = [
'image_id', 'image_category', 'xmin', 'ymin', 'xmax', 'ymax',
'score'
]
df_res = pd.DataFrame(record, columns=cloumns)
results.append(df_res)
df_res.to_csv(cls + '_res.csv', index=False)
data = pd.concat(results, axis=0)
data.to_csv('./rcnn_res.csv', index=None)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
txt_name = args.image_path[:-1]
result_txt = open(txt_name + '.txt', 'wb')
vis = False
for directory in os.listdir(args.image_path):
print directory[-4:]
if ((directory[-4:] != '.txt') and
(directory[-4:] != 'json')) and (directory != 'cache'):
img_dir = args.image_path + directory
for img in os.listdir(img_dir):
demo_net(predictor, img_dir + '/' + img, result_txt, vis)
result_txt.close()
def main():
color = cv2.imread(args.img) # read image in b,g,r order
img, scale = resize(color.copy(), 640, 1024)
im_info = np.array([[img.shape[0], img.shape[1], scale]], dtype=np.float32) # (h, w, scale)
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 1, 2) # change to r,g,b order
img = img[np.newaxis, :] # extend to (n, c, h, w)
ctx = mx.gpu(args.gpu)
_, arg_params, aux_params = mx.model.load_checkpoint(args.prefix, args.epoch)
arg_params, aux_params = ch_dev(arg_params, aux_params, ctx)
if 'resnet' in args.prefix:
sym = resnet_50(num_class=2, bn_mom=0.99, bn_global=True, is_train=False)
else:
sym = get_vgg_test(num_classes=2)
arg_params["data"] = mx.nd.array(img, ctx)
arg_params["im_info"] = mx.nd.array(im_info, ctx)
exe = sym.bind(ctx, arg_params, args_grad=None, grad_req="null", aux_states=aux_params)
exe.forward(is_train=False)
output_dict = {name: nd for name, nd in zip(sym.list_outputs(), exe.outputs)}
rois = output_dict['rpn_rois_output'].asnumpy()[:, 1:] # first column is index
scores = output_dict['cls_prob_reshape_output'].asnumpy()[0]
bbox_deltas = output_dict['bbox_pred_reshape_output'].asnumpy()[0]
pred_boxes = bbox_pred(rois, bbox_deltas)
pred_boxes = clip_boxes(pred_boxes, (im_info[0][0], im_info[0][1]))
cls_boxes = pred_boxes[:, 4:8]
cls_scores = scores[:, 1]
keep = np.where(cls_scores >= args.thresh)[0]
cls_boxes = cls_boxes[keep, :]
cls_scores = cls_scores[keep]
dets = np.hstack((cls_boxes, cls_scores[:, np.newaxis])).astype(np.float32)
keep = nms(dets.astype(np.float32), args.nms_thresh)
dets = dets[keep, :]
keep = nest(dets, thresh=args.nest_thresh)
dets = dets[keep, :]
for i in range(dets.shape[0]):
bbox = dets[i, :4]
cv2.rectangle(color, (int(round(bbox[0]/scale)), int(round(bbox[1]/scale))),
(int(round(bbox[2]/scale)), int(round(bbox[3]/scale))), (0, 255, 0), 2)
cv2.imwrite("result.jpg", color)
def test_net(imageset, year, root_path, devkit_path, prefix, epoch, ctx, vis):
# set config
config.TEST.HAS_RPN = True
# set up logger
logger = logging.getLogger()
logger.setLevel(logging.INFO)
# load testing data
voc, roidb = load_gt_roidb(imageset, year, root_path, devkit_path)
test_data = ROIIter(roidb, batch_size=1, shuffle=False, mode='test')
# load model
args, auxs = load_param(prefix, epoch, convert=True, ctx=ctx)
# load symbol
sym = get_vgg_test()
# detect
detector = Detector(sym, ctx, args, auxs)
pred_eval(detector, test_data, voc, vis=vis)
def test_rcnn(imageset, year, root_path, devkit_path, prefix, epoch, ctx, vis=False, has_rpn=True, proposal='rpn'):
# load symbol and testing data
if has_rpn:
sym = get_vgg_test()
config.TEST.HAS_RPN = True
config.TEST.RPN_PRE_NMS_TOP_N = 6000
config.TEST.RPN_POST_NMS_TOP_N = 300
voc, roidb = load_gt_roidb(imageset, year, root_path, devkit_path)
else:
sym = get_vgg_rcnn_test()
voc, roidb = eval('load_test_' + proposal + '_roidb')(imageset, year, root_path, devkit_path)
# get test data iter
test_data = ROIIter(roidb, batch_size=1, shuffle=False, mode='test')
# load model
args, auxs, _ = load_param(prefix, epoch, convert=True, ctx=ctx)
# detect
detector = Detector(sym, ctx, args, auxs)
pred_eval(detector, test_data, voc, vis=vis)
def test_net(imageset, year, root_path, devkit_path, prefix, epoch, ctx, vis):
# set config
config.TEST.HAS_RPN = True
# set up logger
logger = logging.getLogger()
logger.setLevel(logging.INFO)
# load testing data
voc, roidb = load_gt_roidb(imageset, year, root_path, devkit_path)
test_data = ROIIter(roidb, batch_size=1, shuffle=False, mode='test')
# load model
args, auxs = load_param(prefix, epoch, convert=True, ctx=ctx)
# load symbol
sym = get_vgg_test()
# detect
detector = Detector(sym, ctx, args, auxs)
pred_eval(detector, test_data, voc, vis=vis)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES,
num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
if args.image:
# single test image
demo_net(predictor, args.image, args.vis)
else:
# a image dir for test
# import pdb
img_list = os.listdir(args.dir)
num = len(img_list)
for line in img_list:
img_path = os.path.join(args.dir, line)
if os.path.isfile(img_path) and os.path.splitext(img_path)[-1] in [
'.jpg'
]:
# pdb.set_trace()
logger.info('%s' % num)
demo_net(predictor, img_path, args.vis)
num = num - 1
def get_net(prefix, epoch, ctx):
args, auxs = load_param(prefix, epoch, convert=True, ctx=ctx)
sym = get_vgg_test()
detector = Detector(sym, ctx, args, auxs)
return detector
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=config.NUM_CLASSES, num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
demo_net(predictor, args.image, args.vis)
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test()
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
demo_net(predictor, args.image)
def get_net(prefix, epoch, ctx):
args, auxs, num_class = load_param(prefix, epoch, convert=True, ctx=ctx)
sym = get_vgg_test(num_classes=num_class)
detector = Detector(sym, ctx, args, auxs)
return detector
def main():
args = parse_args()
ctx = mx.gpu(args.gpu)
symbol = get_vgg_test(num_classes=21, num_anchors=9)
predictor = get_net(symbol, args.prefix, args.epoch, ctx)
demo_net(predictor, args.image, args.vis)
help='saved model prefix',
default='model/e2e',
type=str)
parser.add_argument('--epoch',
help='epoch of pretrained model',
default=50,
type=int)
parser.add_argument('--gpu', help='GPU device to use', default=0, type=int)
parser.add_argument('--vis', help='display result', action='store_true')
args = parser.parse_args()
return args
if __name__ == '__main__':
args = parse_args()
ctx = mx.gpu(0)
if (Global.prefix_value == 'model/e2e'):
symbol = get_resnet_test(num_classes=Global.num_class_value,
num_anchors=config.NUM_ANCHORS)
if (Global.prefix_value == 'model/final'):
symbol = get_vgg_test(num_classes=Global.num_class_value,
num_anchors=config.NUM_ANCHORS)
predictor = get_net(symbol, args.prefix, ctx)
from glob import glob
res = glob("/root/mx-rcnn/testimage/bird/*.JPG")
cnt = 0
for i in res:
demo_net(predictor, i, False)
# if cnt>10:
# break