def _work(process_id, infer_dataset, args):
databin = infer_dataset[process_id]
infer_data_loader = DataLoader(databin, shuffle=False, num_workers=0, pin_memory=False)
for iter, pack in enumerate(infer_data_loader):
img_name = voc12.dataloader.decode_int_filename(pack['name'][0])
img = pack['img'][0].numpy()
cam_dict = np.load(os.path.join(args.cam_out_dir, img_name + '.npy'), allow_pickle=True).item()
cams = cam_dict['high_res']
keys = np.pad(cam_dict['keys'] + 1, (1, 0), mode='constant')
# 1. find confident fg & bg
fg_conf_cam = np.pad(cams, ((1, 0), (0, 0), (0, 0)), mode='constant', constant_values=args.conf_fg_thres)
fg_conf_cam = np.argmax(fg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img, fg_conf_cam, n_labels=keys.shape[0])
fg_conf = keys[pred]
bg_conf_cam = np.pad(cams, ((1, 0), (0, 0), (0, 0)), mode='constant', constant_values=args.conf_bg_thres)
bg_conf_cam = np.argmax(bg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img, bg_conf_cam, n_labels=keys.shape[0])
bg_conf = keys[pred]
# 2. combine confident fg & bg
conf = fg_conf.copy()
conf[fg_conf == 0] = 255
conf[bg_conf + fg_conf == 0] = 0
imageio.imwrite(os.path.join(args.ir_label_out_dir, img_name + '.png'),
conf.astype(np.uint8))
if process_id == args.num_workers - 1 and iter % (len(databin) // 20) == 0:
print("%d " % ((5 * iter + 1) // (len(databin) // 20)), end='')
def _work(process_id, infer_dataset, args):
databin = infer_dataset[process_id]
infer_data_loader = DataLoader(databin,
shuffle=False,
num_workers=0,
pin_memory=False)
with tqdm(total=len(infer_data_loader)) as pbar:
for iter, pack in enumerate(infer_data_loader):
img_name = pack['name'][
0] # voc12.dataloader.decode_int_filename(pack['name'][0])
img = pack['img'][0].numpy()
cam_dict = np.load(os.path.join(args.cam_out_dir,
img_name + '.npy'),
allow_pickle=True).item()
if args.dataset in ['adp_morph', 'adp_func']:
keys = np.concatenate((np.array([-1]), cam_dict['keys']))
# 1. find confident fg
fg_conf_cam = np.pad(cam_dict['high_res'],
((1, 0), (0, 0), (0, 0)),
mode='constant',
constant_values=args.conf_fg_thres)
fg_conf_cam = np.argmax(fg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img,
fg_conf_cam,
args.dataset,
n_labels=keys.shape[0])
fg_conf = keys[pred]
# 2. combine confident fg & bg
conf = fg_conf.copy()
conf[fg_conf == -1] = 255
elif args.dataset == 'voc12':
keys = np.pad(cam_dict['keys'] + 1, (1, 0), mode='constant')
# 1. find confident fg & bg
fg_conf_cam = np.pad(cam_dict['high_res'],
((1, 0), (0, 0), (0, 0)),
mode='constant',
constant_values=args.conf_fg_thres)
fg_conf_cam = np.argmax(fg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img,
fg_conf_cam,
args.dataset,
n_labels=keys.shape[0])
fg_conf = keys[pred]
bg_conf_cam = np.pad(cam_dict['high_res'],
((1, 0), (0, 0), (0, 0)),
mode='constant',
constant_values=args.conf_bg_thres)
bg_conf_cam = np.argmax(bg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img,
bg_conf_cam,
args.dataset,
n_labels=keys.shape[0])
bg_conf = keys[pred]
# 2. combine confident fg & bg
conf = fg_conf.copy()
conf[fg_conf == 0] = 255
conf[bg_conf + fg_conf == 0] = 0
elif args.dataset in ['deepglobe', 'deepglobe_balanced']:
keys = np.concatenate((np.array([-1]), cam_dict['keys']))
img = cv2.resize(img, (img.shape[0] // 4, img.shape[1] // 4))
# 1. find confident fg
fg_conf_cam = np.pad(cam_dict['cam'], ((1, 0), (0, 0), (0, 0)),
mode='constant',
constant_values=args.conf_fg_thres)
fg_conf_cam = np.argmax(fg_conf_cam, axis=0)
pred = imutils.crf_inference_label(img,
fg_conf_cam,
args.dataset,
n_labels=keys.shape[0])
fg_conf = keys[pred]
# 2. combine confident fg & bg
conf = fg_conf.copy()
conf[fg_conf == -1] = 255
else:
raise KeyError('Dataset %s not yet implemented' % args.dataset)
imageio.imwrite(
os.path.join(args.ir_label_out_dir, img_name + '.png'),
conf.astype(np.uint8))
# Save with colour
cam_clr = np.zeros(list(conf.shape) + [3], dtype=np.uint8)
off = 0
for t in ['bg', 'fg']:
for i, c in enumerate(args.class_colours[t]):
for ch in range(3):
cam_clr[:, :,
ch] += c[ch] * np.uint8(conf == (i + off))
off += len(args.class_colours[t])
for ch in range(3):
cam_clr[:, :, ch] += 255 * np.uint8(conf == 255)
imageio.imsave(
os.path.join(args.ir_label_clr_out_dir, img_name + '.png'),
cam_clr)
# Save with colour, overlaid on original image
cam_clr_over = np.uint8((1 - args.overlay_r) * np.float32(img) +
args.overlay_r * np.float32(cam_clr))
imageio.imsave(
os.path.join(args.ir_label_clr_out_dir,
img_name + '_overlay.png'), cam_clr_over)
pbar.update(1)