'samples': [samples[i]]
}]
blob = get_next_mini_batch(db)
blob = {'data': blob['data']}
net.blobs['data'].reshape(*blob['data'].shape)
out = net.forward(**blob)['cls_prob']
scores[i] = out[1]
if __name__ == '__main__':
caffe.set_mode_gpu()
caffe.set_device(1)
# get the deploy solver and net with pre-trained caffe model
train = os.path.join('model', 'deploy_solver.prototxt')
test = os.path.join('model', 'deploy_test.prototxt')
weights = os.path.join('model', 'MDNet_iter_800000.caffemodel')
solver, net = get_solver_net(train, test, weights)
# get the Evaluator
dtype = 'VOT'
dbpath = os.path.join('data', 'vot2014')
gtpath = dbpath
vdbc = VDBC(dbtype=dtype, dbpath=dbpath, gtpath=gtpath, flush=True)
evl = Evaluator(vdbc)
evl.set_video(1)
evaluate(evl, solver, net)
if score < threshold:
finetune(solver, frame_samples, short_term)
elif term % 10 == 0:
finetune(solver, frame_samples, long_term)
evl.report(box.reshape((4, )))
gt = box.reshape((4, ))
if VISUAL:
ground_truth = evl.get_ground_truth()
vis_detection(im_path, ground_truth, gt)
im_path = evl.next_frame()
if __name__ == '__main__':
solver, net = get_solver_net(train, test, weights)
# get the Evaluator
dtype = 'VOT'
dbpath = os.path.join('data', 'vot2014')
gtpath = dbpath
vdbc = VDBC(dbtype=dtype, dbpath=dbpath, gtpath=gtpath, flush=True)
evl = Evaluator(vdbc)
video_num = evl.get_video_num()
print 'Total video sequences: {}.'.format(video_num)
for i in range(video_num):
evaluate(evl, solver, net)
plt.gca().add_patch(
plt.Rectangle((box[0], box[1]),
box[2],
box[3],
fill=False,
edgecolor='blue',
linewidth=1.5))
plt.show()
if __name__ == '__main__':
IM_PER_FRAME = 256
dtype = 'VOT'
dbpath = os.path.join('data', 'VOT')
gtpath = dbpath
vdbc = VDBC(dbtype=dtype, dbpath=dbpath, gtpath=gtpath, flush=True)
evl = Evaluator(vdbc)
evl.set_video(3)
im_path, gt = evl.init_frame()
im = cv2.imread(im_path)
frame_samples = mdnet_sample(im, gt, PARAMS, IM_PER_FRAME, 'TEST')
#frame_samples = uniform_sample(im ,gt, PARAMS, IM_PER_FRAME, 'TEST')
bboxes = [sample['box'] for sample in frame_samples]
vis_detection(im_path, gt, bboxes)
#print type(bboxes)
print 'mAP: {}.'.format(evl.get_mAP())
record.add_record(sample_num=sample_num,
frame_num=term,
mAP=evl.get_mAP(),
total_time=total_timer.diff,
finetune_iter=finetune_iter_)
record._save_json()
if __name__ == '__main__':
while IMS_PER_FRAME > 20:
solver, net = get_solver_net(train, test, weights)
# get the Evaluator
dtype = 'VOT'
dbpath = os.path.join('data', 'VOT')
gtpath = dbpath
vdbc = VDBC(dbtype=dtype, dbpath=dbpath, gtpath=gtpath, flush=True)
evl = Evaluator(vdbc)
video_num = evl.get_video_num()
print 'Total video sequences: {}.'.format(video_num)
# for i in range(video_num):
# evaluate(evl, solver, net)
evl.set_video(19)
evaluate(evl, solver, net, IMS_PER_FRAME)
IMS_PER_FRAME -= 5