def sample(model, model_path, output_name='gen/gen'):
pred = PredictConfig(session_init=get_model_loader(model_path),
model=model,
input_names=['z'],
output_names=[output_name, 'z'])
pred = SimpleDatasetPredictor(pred, RandomZData((100, args.z_dim)))
for o in pred.get_result():
o = o[0] + 1
o = o * 128.0
o = np.clip(o, 0, 255)
o = o[:, :, :, ::-1]
stack_patches(o, nr_row=10, nr_col=10, viz=True)
def sample(model, model_path, output_name='gen/gen'):
pred = PredictConfig(
session_init=get_model_loader(model_path),
model=model,
input_names=['z'],
output_names=[output_name, 'z'])
pred = SimpleDatasetPredictor(pred, RandomZData((100, args.z_dim)))
for o in pred.get_result():
o = o[0] + 1
o = o * 128.0
o = np.clip(o, 0, 255)
o = o[:, :, :, ::-1]
stack_patches(o, nr_row=10, nr_col=10, viz=True)
def sample(datadir, model_path):
pred = PredictConfig(session_init=get_model_loader(model_path),
model=Model(),
input_names=['input', 'output'],
output_names=['viz'])
imgs = glob.glob(os.path.join(datadir, '*.jpg'))
ds = ImageFromFile(imgs, channel=3, shuffle=True)
ds = MapData(ds, lambda dp: split_input(dp[0]))
ds = AugmentImageComponents(ds, [imgaug.Resize(256)], (0, 1))
ds = BatchData(ds, 6)
pred = SimpleDatasetPredictor(pred, ds)
for o in pred.get_result():
o = o[0][:, :, :, ::-1]
stack_patches(o, nr_row=3, nr_col=2, viz=True)
def infer_folder(datadir, model_path):
pred = PredictConfig(
session_init=get_model_loader(model_path),
model=Model(),
input_names=['input', 'output'],
output_names=['viz'])
imgs = glob.glob(os.path.join(datadir, '*' + IMG_TYPE))
ds = CustomDataFlow(datadir, 'test', shuffle=True)
ds = AugmentImageComponents(ds, [imgaug.Resize(256)], (0, 1))
ds = BatchData(ds, 6)
pred = SimpleDatasetPredictor(pred, ds)
for o in pred.get_result():
o = o[0][:, :, :, ::-1]
stack_patches(o, nr_row=3, nr_col=2, viz=True)
def visualize(model, model_path, nr_visualize=100, output_dir='output'):
"""
Visualize some intermediate results (proposals, raw predictions) inside the pipeline.
"""
df = get_train_dataflow() # we don't visualize mask stuff
df.reset_state()
pred = OfflinePredictor(
PredictConfig(model=model,
session_init=get_model_loader(model_path),
input_names=['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_{}_proposals/boxes'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'generate_{}_proposals/probs'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'fastrcnn_all_probs',
'final_boxes',
'final_probs',
'final_labels',
]))
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
utils.fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df.get_data()),
nr_visualize):
img = dp[0]
if cfg.MODE_MASK:
gt_boxes, gt_labels, gt_masks = dp[-3:]
else:
gt_boxes, gt_labels = dp[-2:]
rpn_boxes, rpn_scores, all_probs, \
final_boxes, final_probs, final_labels = pred(img, gt_boxes, gt_labels)
# draw groundtruth boxes
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
# draw best proposals for each groundtruth, to show recall
proposal_viz, good_proposals_ind = draw_proposal_recall(
img, rpn_boxes, rpn_scores, gt_boxes)
# draw the scores for the above proposals
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind],
all_probs[good_proposals_ind])
results = [
DetectionResult(*args)
for args in zip(final_boxes, final_probs, final_labels,
[None] * len(final_labels))
]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches(
[gt_viz, proposal_viz, score_viz, final_viz], 2, 2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite("{}/{:03d}.png".format(output_dir, idx), viz)
pbar.update()
def apply(model, model_path, images, ground_truth=None):
left = cv2.imread(images[0])
h, w = left.shape[:2]
newh = (h // 64) * 64
neww = (w // 64) * 64
aug = imgaug.CenterCrop((newh, neww))
left = aug.augment(left)
predict_func = OfflinePredictor(
PredictConfig(model=model(height=newh, width=neww),
session_init=get_model_loader(model_path),
input_names=['left', 'right'],
output_names=['prediction']))
for right in images[1:]:
right = aug.augment(cv2.imread(right))
left_input, right_input = [
x.astype('float32').transpose(2, 0, 1)[None, ...]
for x in [left, right]
]
output = predict_func(left_input, right_input)[0].transpose(0, 2, 3, 1)
flow = Flow()
img = flow.visualize(output[0])
patches = [left, right, img * 255.]
if ground_truth is not None:
patches.append(flow.visualize(Flow.read(ground_truth)) * 255.)
img = viz.stack_patches(patches, 2, 2)
cv2.imshow('flow output', img)
cv2.imwrite('flow_prediction.png', img)
cv2.waitKey(0)
left = right
def sample(datadir, model_path):
pred = PredictConfig(
session_init=get_model_loader(model_path),
model=Model(),
input_names=['input', 'output'],
output_names=['viz'])
imgs = glob.glob(os.path.join(datadir, '*.jpg'))
ds = ImageFromFile(imgs, channel=3, shuffle=True)
ds = MapData(ds, lambda dp: split_input(dp[0]))
ds = AugmentImageComponents(ds, [imgaug.Resize(256)], (0, 1))
ds = BatchData(ds, 6)
pred = SimpleDatasetPredictor(pred, ds)
for o in pred.get_result():
o = o[0][:, :, :, ::-1]
stack_patches(o, nr_row=3, nr_col=2, viz=True)
def visualize(model_path, nr_visualize=50, output_dir='output'):
df = get_train_dataflow_coco() # we don't visualize mask stuff
df.reset_state()
pred = OfflinePredictor(
PredictConfig(model=Model(),
session_init=get_model_loader(model_path),
input_names=['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_rpn_proposals/boxes',
'generate_rpn_proposals/probs',
'fastrcnn_all_probs',
'final_boxes',
'final_probs',
'final_labels',
]))
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
utils.fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df.get_data()),
nr_visualize):
img, _, _, gt_boxes, gt_labels = dp
rpn_boxes, rpn_scores, all_probs, \
final_boxes, final_probs, final_labels = pred(img, gt_boxes, gt_labels)
# draw groundtruth boxes
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
# draw best proposals for each groundtruth, to show recall
proposal_viz, good_proposals_ind = draw_proposal_recall(
img, rpn_boxes, rpn_scores, gt_boxes)
# draw the scores for the above proposals
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind],
all_probs[good_proposals_ind])
if config.USE_SECOND_HEAD:
results = [
SecondDetectionResult(*args)
for args in zip(final_boxes, final_probs, final_labels,
[None] * len(final_labels))
]
else:
results = [
DetectionResult(*args)
for args in zip(final_boxes, final_probs, final_labels,
[None] * len(final_labels))
]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches(
[gt_viz, proposal_viz, score_viz, final_viz], 2, 2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite("{}/{:03d}.png".format(output_dir, idx), viz)
pbar.update()
def sample(model_path, output_name, BATCH_SIZE, Net, res):
#output_name = 'gen/toRGBtemp_128_256/output'
BATCH_SIZE = int(BATCH_SIZE)
pred = PredictConfig(
session_init=get_model_loader(model_path),
model=Net.Model(),
input_names=['z'],
output_names=[output_name, 'z'])
inputNoise = RandomZData((BATCH_SIZE, 512))
#inputNoise = ZData((BATCH_SIZE, 512))
pred = SimpleDatasetPredictor(pred, inputNoise)
its = 0
for o in pred.get_result():
o = o[0] + 1
o = o * 128.0
o = np.clip(o, 0, 255)
o = o[:, :, :, ::-1]
stack_patches(o, nr_row=10, nr_col=10, viz=True)
def sample(model_path):
pred = OfflinePredictor(
PredictConfig(session_init=get_model_loader(model_path),
model=Model(),
input_names=['z_code', 'z_noise'],
output_names=['gen/viz']))
# sample all one-hot encodings (10 times)
z_cat = np.tile(np.eye(10), [10, 1])
# sample continuos variables from -2 to +2 as mentioned in the paper
z_uni = np.linspace(-2.0, 2.0, num=100)
z_uni = z_uni[:, None]
IMG_SIZE = 400
while True:
# only categorical turned on
z_noise = np.random.uniform(-1, 1, (100, NOISE_DIM))
zc = np.concatenate((z_cat, z_uni * 0, z_uni * 0), axis=1)
o = pred(zc, z_noise)[0]
viz1 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz1 = cv2.resize(viz1, (IMG_SIZE, IMG_SIZE))
# show effect of first continous variable with fixed noise
zc = np.concatenate((z_cat, z_uni, z_uni * 0), axis=1)
o = pred(zc, z_noise * 0)[0]
viz2 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz2 = cv2.resize(viz2, (IMG_SIZE, IMG_SIZE))
# show effect of second continous variable with fixed noise
zc = np.concatenate((z_cat, z_uni * 0, z_uni), axis=1)
o = pred(zc, z_noise * 0)[0]
viz3 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz3 = cv2.resize(viz3, (IMG_SIZE, IMG_SIZE))
canvas = viz.stack_patches([viz1, viz2, viz3],
nr_row=1,
nr_col=3,
border=5,
bgcolor=(255, 0, 0))
viz.interactive_imshow(canvas)
def visualize(model, model_path, nr_visualize=100, output_dir='output'):
"""
Visualize some intermediate results (proposals, raw predictions) inside the pipeline.
"""
df = get_train_dataflow() # we don't visualize mask stuff
df.reset_state()
pred = OfflinePredictor(PredictConfig(
model=model,
session_init=get_model_loader(model_path),
input_names=['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_{}_proposals/boxes'.format('fpn' if cfg.MODE_FPN else 'rpn'),
'generate_{}_proposals/scores'.format('fpn' if cfg.MODE_FPN else 'rpn'),
'fastrcnn_all_scores',
'output/boxes',
'output/scores',
'output/labels',
]))
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
utils.fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df), nr_visualize):
img = dp[0]
if cfg.MODE_MASK:
gt_boxes, gt_labels, gt_masks = dp[-3:]
else:
gt_boxes, gt_labels = dp[-2:]
rpn_boxes, rpn_scores, all_scores, \
final_boxes, final_scores, final_labels = pred(img, gt_boxes, gt_labels)
# draw groundtruth boxes
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
# draw best proposals for each groundtruth, to show recall
proposal_viz, good_proposals_ind = draw_proposal_recall(img, rpn_boxes, rpn_scores, gt_boxes)
# draw the scores for the above proposals
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind], all_scores[good_proposals_ind])
results = [DetectionResult(*args) for args in
zip(final_boxes, final_scores, final_labels,
[None] * len(final_labels))]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches([
gt_viz, proposal_viz,
score_viz, final_viz], 2, 2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite("{}/{:03d}.png".format(output_dir, idx), viz)
pbar.update()
def sample(model_path):
pred = OfflinePredictor(PredictConfig(
session_init=get_model_loader(model_path),
model=Model(),
input_names=['z_code', 'z_noise'],
output_names=['gen/viz']))
# sample all one-hot encodings (10 times)
z_cat = np.tile(np.eye(10), [10, 1])
# sample continuos variables from -2 to +2 as mentioned in the paper
z_uni = np.linspace(-2.0, 2.0, num=100)
z_uni = z_uni[:, None]
IMG_SIZE = 400
while True:
# only categorical turned on
z_noise = np.random.uniform(-1, 1, (100, NOISE_DIM))
zc = np.concatenate((z_cat, z_uni * 0, z_uni * 0), axis=1)
o = pred(zc, z_noise)[0]
viz1 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz1 = cv2.resize(viz1, (IMG_SIZE, IMG_SIZE))
# show effect of first continous variable with fixed noise
zc = np.concatenate((z_cat, z_uni, z_uni * 0), axis=1)
o = pred(zc, z_noise * 0)[0]
viz2 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz2 = cv2.resize(viz2, (IMG_SIZE, IMG_SIZE))
# show effect of second continous variable with fixed noise
zc = np.concatenate((z_cat, z_uni * 0, z_uni), axis=1)
o = pred(zc, z_noise * 0)[0]
viz3 = viz.stack_patches(o, nr_row=10, nr_col=10)
viz3 = cv2.resize(viz3, (IMG_SIZE, IMG_SIZE))
canvas = viz.stack_patches(
[viz1, viz2, viz3],
nr_row=1, nr_col=3, border=5, bgcolor=(255, 0, 0))
viz.interactive_imshow(canvas)
def sample(model_path):
pred = PredictConfig(session_init=get_model_loader(model_path),
model=Model(),
input_names=['label', 'z'],
output_names=['gen/gen'])
ds = MapData(RandomZData((100, 100)),
lambda dp: [np.arange(100) % 10, dp[0]])
pred = SimpleDatasetPredictor(pred, ds)
for o in pred.get_result():
o = o[0] * 255.0
viz = stack_patches(o, nr_row=10, nr_col=10)
viz = cv2.resize(viz, (800, 800))
interactive_imshow(viz)
def sample(model_path):
pred = PredictConfig(
session_init=get_model_loader(model_path),
model=Model(),
input_names=['label', 'z'],
output_names=['gen/gen'])
ds = MapData(RandomZData((100, 100)),
lambda dp: [np.arange(100) % 10, dp[0]])
pred = SimpleDatasetPredictor(pred, ds)
for o in pred.get_result():
o = o[0] * 255.0
viz = stack_patches(o, nr_row=10, nr_col=10)
viz = cv2.resize(viz, (800, 800))
interactive_imshow(viz)
def visualize(model_path, nr_visualize=50, output_dir='output'):
pred = OfflinePredictor(
PredictConfig(model=Model(),
session_init=get_model_loader(model_path),
input_names=['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_rpn_proposals/boxes',
'generate_rpn_proposals/probs',
'fastrcnn_all_probs',
'fastrcnn_fg_probs',
'fastrcnn_fg_boxes',
]))
df = get_train_dataflow()
df.reset_state()
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
utils.fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df.get_data()),
nr_visualize):
img, _, _, gt_boxes, gt_labels = dp
rpn_boxes, rpn_scores, all_probs, fg_probs, fg_boxes = pred(
img, gt_boxes, gt_labels)
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
proposal_viz, good_proposals_ind = draw_proposal_recall(
img, rpn_boxes, rpn_scores, gt_boxes)
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind],
all_probs[good_proposals_ind])
fg_boxes = clip_boxes(fg_boxes, img.shape[:2])
fg_viz = draw_predictions(img, fg_boxes, fg_probs)
results = nms_fastrcnn_results(fg_boxes, fg_probs)
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches(
[gt_viz, proposal_viz, score_viz, fg_viz, final_viz], 2, 3)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite("{}/{:03d}.png".format(output_dir, idx), viz)
pbar.update()
def predict_unlabeled(model,
model_path,
nr_visualize=100,
output_dir='output_patch_samples'):
"""Predict the pseudo label information of unlabeled data."""
assert cfg.EVAL.PSEUDO_INFERENCE, 'set cfg.EVAL.PSEUDO_INFERENCE=True'
df, dataset_size = get_eval_unlabeled_dataflow(cfg.DATA.TRAIN,
return_size=True)
df.reset_state()
predcfg = PredictConfig(
model=model,
session_init=SmartInit(model_path),
input_names=['image'], # ['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_{}_proposals/boxes'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'generate_{}_proposals/scores'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'fastrcnn_all_scores',
'output/boxes',
'output/scores', # score of the labels
'output/labels',
])
pred = OfflinePredictor(predcfg)
if os.path.isdir(output_dir):
if os.path.isfile(os.path.join(output_dir, 'pseudo_data.npy')):
os.remove(os.path.join(output_dir, 'pseudo_data.npy'))
if not os.path.isdir(os.path.join(output_dir, 'vis')):
os.makedirs(os.path.join(output_dir, 'vis'))
else:
shutil.rmtree(os.path.join(output_dir, 'vis'))
fs.mkdir_p(output_dir + '/vis')
else:
fs.mkdir_p(output_dir)
fs.mkdir_p(output_dir + '/vis')
logger.warning('-' * 100)
logger.warning('Write to {}'.format(output_dir))
logger.warning('-' * 100)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df), nr_visualize):
img, img_id = dp # dp['image'], dp['img_id']
rpn_boxes, rpn_scores, all_scores, \
final_boxes, final_scores, final_labels = pred(img)
outs = {
'proposals_boxes': rpn_boxes, # (?,4)
'proposals_scores': rpn_scores, # (?,)
'boxes': final_boxes,
'scores': final_scores,
'labels': final_labels
}
ratios = [10,
10] # [top 20% as background, bottom 20% as background]
bg_ind, fg_ind = custom.find_bg_and_fg_proposals(all_scores,
ratios=ratios)
bg_viz = draw_predictions(img, rpn_boxes[bg_ind],
all_scores[bg_ind])
fg_viz = draw_predictions(img, rpn_boxes[fg_ind],
all_scores[fg_ind])
results = [
DetectionResult(*args)
for args in zip(final_boxes, final_scores, final_labels,
[None] * len(final_labels))
]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches([bg_viz, fg_viz, final_viz], 2, 2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
assert cv2.imwrite('{}/vis/{:03d}.png'.format(output_dir, idx),
viz)
pbar.update()
logger.info('Write {} samples to {}'.format(nr_visualize, output_dir))
## Parallel inference the whole unlabled data
pseudo_preds = collections.defaultdict(list)
num_tower = max(cfg.TRAIN.NUM_GPUS, 1)
graph_funcs = MultiTowerOfflinePredictor(predcfg, list(
range(num_tower))).get_predictors()
dataflows = [
get_eval_unlabeled_dataflow(cfg.DATA.TRAIN,
shard=k,
num_shards=num_tower)
for k in range(num_tower)
]
all_results = multithread_predict_dataflow(dataflows, graph_funcs)
for id, result in tqdm.tqdm(enumerate(all_results)):
img_id = result['image_id']
outs = {
'proposals_boxes':
result['proposal_box'].astype(np.float16), # (?,4)
'proposals_scores':
result['proposal_score'].astype(np.float16), # (?,)
# 'frcnn_all_scores': result['frcnn_score'].astype(np.float16),
'boxes': result['bbox'].astype(np.float16), # (?,4)
'scores': result['score'].astype(np.float16), # (?,)
'labels': result['category_id'].astype(np.float16) # (?,)
}
pseudo_preds[img_id] = outs
logger.warn('Writing to {}'.format(
os.path.join(output_dir, 'pseudo_data.npy')))
try:
dd.io.save(os.path.join(output_dir, 'pseudo_data.npy'), pseudo_preds)
except RuntimeError:
logger.error('Save failed. Check reasons manually...')
def collect_samples(model,
model_path,
nr_visualize=100,
output_dir='output_patch_samples'):
"""
Visualize some intermediate results (proposals, raw predictions) inside the
pipeline.
"""
df = get_train_dataflow()
df.reset_state()
pred = OfflinePredictor(
PredictConfig(model=model,
session_init=SmartInit(model_path),
input_names=['image', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_{}_proposals/boxes'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'generate_{}_proposals/scores'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'fastrcnn_all_scores',
'output/boxes',
'output/scores',
'output/labels',
]))
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df), nr_visualize):
img, gt_boxes, gt_labels = dp['image'], dp['gt_boxes'], dp[
'gt_labels']
rpn_boxes, rpn_scores, all_scores, \
final_boxes, final_scores, final_labels = pred(img, gt_boxes, gt_labels)
# draw groundtruth boxes
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
ratios = [10,
10] #[top 20% as background, bottom 20% as background]
bg_ind, fg_ind = custom.find_bg_and_fg_proposals(all_scores,
ratios=ratios)
bg_viz = draw_predictions(img, rpn_boxes[bg_ind],
all_scores[bg_ind])
fg_viz = draw_predictions(img, rpn_boxes[fg_ind],
all_scores[fg_ind])
# draw best proposals for each groundtruth, to show recall
proposal_viz, good_proposals_ind = draw_proposal_recall(
img, rpn_boxes, rpn_scores, gt_boxes)
# draw the scores for the above proposals
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind],
all_scores[good_proposals_ind])
results = [
DetectionResult(*args)
for args in zip(final_boxes, final_scores, final_labels,
[None] * len(final_labels))
]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches([gt_viz, bg_viz, fg_viz, final_viz], 2,
2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite('{}/{:03d}.png'.format(output_dir, idx), viz)
pbar.update()
def do_visualize(model, model_path, nr_visualize=100, output_dir='output'):
"""
Visualize some intermediate results (proposals, raw predictions) inside the pipeline.
"""
df = get_train_dataflow() # we don't visualize mask stuff
df.reset_state()
pred = OfflinePredictor(
PredictConfig(
model=model,
session_init=get_model_loader(model_path),
input_names=['images', 'orig_image_dims', 'gt_boxes', 'gt_labels'],
output_names=[
'generate_{}_proposals_topk_per_image/boxes'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'generate_{}_proposals_topk_per_image/scores'.format(
'fpn' if cfg.MODE_FPN else 'rpn'),
'fastrcnn_all_scores',
'output/boxes',
'output/scores',
'output/labels',
]))
if os.path.isdir(output_dir):
shutil.rmtree(output_dir)
utils.fs.mkdir_p(output_dir)
with tqdm.tqdm(total=nr_visualize) as pbar:
for idx, dp in itertools.islice(enumerate(df), nr_visualize):
img, gt_boxes, gt_labels = dp['images'], dp['gt_boxes'], dp[
'gt_labels']
orig_shape = img.shape[:2]
rpn_boxes, rpn_scores, all_scores, \
final_boxes, final_scores, final_labels = pred(np.expand_dims(img, axis=0),
np.expand_dims(np.array(img.shape), axis=0),
np.expand_dims(gt_boxes, axis=0),
np.expand_dims(gt_labels, axis=0))
# draw groundtruth boxes
gt_viz = draw_annotation(img, gt_boxes, gt_labels)
# draw best proposals for each groundtruth, to show recall
# custom op creates different shape for boxes, convert back to original
rpn_boxes = np.array([i[1:] for i in rpn_boxes])
proposal_viz, good_proposals_ind = draw_proposal_recall(
img, rpn_boxes, rpn_scores, gt_boxes)
# draw the scores for the above proposals
score_viz = draw_predictions(img, rpn_boxes[good_proposals_ind],
all_scores[good_proposals_ind])
results = [
DetectionResult(*args)
for args in zip(final_boxes, final_scores, final_labels,
[None] * len(final_labels))
]
final_viz = draw_final_outputs(img, results)
viz = tpviz.stack_patches(
[gt_viz, proposal_viz, score_viz, final_viz], 2, 2)
if os.environ.get('DISPLAY', None):
tpviz.interactive_imshow(viz)
cv2.imwrite("{}/{:03d}.png".format(output_dir, idx), viz)
pbar.update()