def fpn_map_rois_to_levels(boxes):
"""
Assign boxes to level 2~5.
Args:
boxes (nx4):
Returns:
[tf.Tensor]: 4 tensors for level 2-5. Each tensor is a vector of indices of boxes in its level.
[tf.Tensor]: 4 tensors, the gathered boxes in each level.
Be careful that the returned tensor could be empty.
"""
sqrtarea = tf.sqrt(tf_area(boxes))
level = tf.to_int32(tf.floor(
4 + tf.log(sqrtarea * (1. / 224) + 1e-6) * (1.0 / np.log(2))))
# RoI levels range from 2~5 (not 6)
level_ids = [
tf.where(level <= 2),
tf.where(tf.equal(level, 3)), # == is not supported
tf.where(tf.equal(level, 4)),
tf.where(level >= 5)]
level_ids = [tf.reshape(x, [-1], name='roi_level{}_id'.format(i + 2))
for i, x in enumerate(level_ids)]
num_in_levels = [tf.size(x, name='num_roi_level{}'.format(i + 2))
for i, x in enumerate(level_ids)]
add_moving_summary(*num_in_levels)
level_boxes = [tf.gather(boxes, ids) for ids in level_ids]
return level_ids, level_boxes
def fpn_map_rois_to_levels(boxes):
"""
Assign boxes to level 2~5.
Args:
boxes (nx4):
Returns:
[tf.Tensor]: 4 tensors for level 2-5. Each tensor is a vector of indices of boxes in its level.
[tf.Tensor]: 4 tensors, the gathered boxes in each level.
Be careful that the returned tensor could be empty.
"""
sqrtarea = tf.sqrt(tf_area(boxes))
level = tf.cast(tf.floor(
4 + tf.math.log(sqrtarea * (1. / 224) + 1e-6) * (1.0 / np.log(2))), tf.int32)
# RoI levels range from 2~5 (not 6)
level_ids = [
tf.where(level <= 2),
tf.where(tf.equal(level, 3)), # == is not supported
tf.where(tf.equal(level, 4)),
tf.where(level >= 5)]
level_ids = [tf.reshape(x, [-1], name='roi_level{}_id'.format(i + 2))
for i, x in enumerate(level_ids)]
num_in_levels = [tf.size(x, name='num_roi_level{}'.format(i + 2))
for i, x in enumerate(level_ids)]
add_moving_summary(*num_in_levels)
level_boxes = [tf.gather(boxes, ids) for ids in level_ids]
return level_ids, level_boxes
def run_head(self, proposals, stage):
"""
Args:
proposals: BoxProposals
stage: 0, 1, 2
Returns:
FastRCNNHead
Nx4, updated boxes
"""
reg_weights = tf.constant(cfg.CASCADE.BBOX_REG_WEIGHTS[stage],
dtype=tf.float32)
pooled_feature = self.roi_func(proposals.boxes) # N,C,S,S
pooled_feature = self.scale_gradient(pooled_feature)
head_feature = self.fastrcnn_head_func('head', pooled_feature)
label_logits, box_logits = fastrcnn_outputs(
'outputs',
head_feature,
self.num_categories,
class_agnostic_regression=True)
head = FastRCNNHead(proposals, box_logits, label_logits, self.gt_boxes,
reg_weights)
refined_boxes = head.decoded_output_boxes_class_agnostic()
refined_boxes = clip_boxes(refined_boxes, self.image_shape2d)
if self.training:
refined_boxes = tf.boolean_mask(refined_boxes,
tf_area(refined_boxes) > 0)
return head, tf.stop_gradient(refined_boxes, name='output_boxes')
def build_graph(self, *inputs):
inputs = dict(zip(self.input_names, inputs))
if "gt_masks_packed" in inputs:
gt_masks = tf.cast(unpackbits_masks(inputs.pop("gt_masks_packed")), tf.uint8, name="gt_masks")
inputs["gt_masks"] = gt_masks
image = self.preprocess(inputs['image']) # 1CHW
features = self.backbone(image)
anchor_inputs = {k: v for k, v in inputs.items() if k.startswith('anchor_')}
proposals, rpn_losses = self.rpn(image, features, anchor_inputs) # inputs?
targets = [inputs[k] for k in ['gt_boxes', 'gt_labels', 'gt_masks'] if k in inputs]
gt_boxes_area = tf.reduce_mean(tf_area(inputs["gt_boxes"]), name='mean_gt_box_area')
add_moving_summary(gt_boxes_area)
head_losses = self.roi_heads(image, features, proposals, targets)
if self.training:
wd_cost = regularize_cost(
'.*/W', l2_regularizer(cfg.TRAIN.WEIGHT_DECAY), name='wd_cost')
total_cost = tf.add_n(
rpn_losses + head_losses + [wd_cost], 'total_cost')
add_moving_summary(total_cost, wd_cost)
return total_cost
else:
# Check that the model defines the tensors it declares for inference
# For existing models, they are defined in "fastrcnn_predictions(name_scope='output')"
G = tf.get_default_graph()
ns = G.get_name_scope()
for name in self.get_inference_tensor_names()[1]:
try:
name = '/'.join([ns, name]) if ns else name
G.get_tensor_by_name(name + ':0')
except KeyError:
raise KeyError("Your model does not define the tensor '{}' in inference context.".format(name))
def fpn_map_rois_to_levels(boxes):
"""
Assign boxes to level 2~5.
Args:
boxes: t x 5, t is the number of sampled boxes
Returns:
level_ids[tf.Tensor]: 4 tensors for level 2-5. Each tensor is a vector of indices of boxes in its level.
level_boxes[tf.Tensor]: 4 tensors, the gathered boxes in each level.
Be careful that the returned tensor could be empty.
"""
sqrtarea = tf.sqrt(tf_area(boxes[:,1:]))
# Map equation from the FPN paper: https://arxiv.org/abs/1612.03144, page 4
# k = [k0 + log2(sqrt(wh)/224)]
level = tf.cast(tf.floor(
4 + tf.log(sqrtarea * (1. / 224) + 1e-6) * (1.0 / np.log(2))), tf.int32)
# RoI levels range from 2~5 (not 6)
level_ids = [
tf.where(level <= 2),
tf.where(tf.equal(level, 3)), # == is not supported
tf.where(tf.equal(level, 4)),
tf.where(level >= 5)]
level_ids = [tf.reshape(x, [-1], name='roi_level{}_id'.format(i + 2))
for i, x in enumerate(level_ids)]
num_in_levels = [tf.size(x, name='num_roi_level{}'.format(i + 2))
for i, x in enumerate(level_ids)]
add_moving_summary(*num_in_levels)
level_boxes = [tf.gather(boxes, ids) for ids in level_ids]
return level_ids, level_boxes
def build_graph(self, *inputs):
print(f"self.input_names : {self.input_names}")
print(F"inputs before : {inputs}")
inputs = dict(zip(self.input_names, inputs))
print(f"inputs : {inputs}")
if "gt_masks_packed" in inputs:
gt_masks = tf.cast(unpackbits_masks(inputs.pop("gt_masks_packed")), tf.uint8, name="gt_masks")
inputs["gt_masks"] = gt_masks
print("inputs[gt_masks] :", inputs["gt_masks"])
# inputs['image'] = tf.Print(inputs['image'], [tf.shape(inputs['image'])], message="image before preprocess : ", summarize=100)
image = self.preprocess(inputs['image']) # 1CHW
# image = tf.Print(image, [tf.shape(image)], message="image after preprocess : ", summarize=100)
features = self.backbone(image)
for i, feature in enumerate(features):
feature = tf.Print(feature, [tf.shape(feature)], message=f"feature p{i+2} : ", summarize=100)
# 여기까지 봄
anchor_inputs = {k: v for k, v in inputs.items() if k.startswith('anchor_')}
# anchor_inputs = tf.Print(anchor_inputs, [anchor_inputs], message="anchor_inputs : ", summarize=100)
proposals, rpn_losses = self.rpn(image, features, anchor_inputs) # inputs?
# proposals = tf.Print(proposals, [tf.shape(proposals)], message="proposals : ", summarize=100)
# rpn_losses = tf.Print(rpn_losses, [tf.shape(rpn_losses)], message="rpn_losses : ", summarize=100)
targets = [inputs[k] for k in ['gt_boxes', 'gt_labels', 'gt_masks'] if k in inputs]
gt_boxes_area = tf.reduce_mean(tf_area(inputs["gt_boxes"]), name='mean_gt_box_area')
add_moving_summary(gt_boxes_area)
head_losses = self.roi_heads(image, features, proposals, targets)
if self.training:
print("l2_regularizer")
print(l2_regularizer)
wd_cost = regularize_cost(
'.*/W', l2_regularizer(cfg.TRAIN.WEIGHT_DECAY), name='wd_cost')
total_cost = tf.add_n(
rpn_losses + head_losses + [wd_cost], 'total_cost')
add_moving_summary(total_cost, wd_cost)
return total_cost
else:
# Check that the model defines the tensors it declares for inference
# For existing models, they are defined in "fastrcnn_predictions(name_scope='output')"
G = tf.get_default_graph()
ns = G.get_name_scope()
for name in self.get_inference_tensor_names()[1]:
try:
name = '/'.join([ns, name]) if ns else name
G.get_tensor_by_name(name + ':0')
except KeyError:
raise KeyError("Your model does not define the tensor '{}' in inference context.".format(name))
