def _build_graph(self, inputs):
is_training = get_current_tower_context().is_training
image, label = inputs
#tf.summary.image('viz', image, max_outputs=10)
image = image_preprocess(image, bgr=False)
#image = image * (1.0 / 255)
image = tf.transpose(image, [0, 3, 1, 2])
depth = config.RESNET_DEPTH
basicblock = preresnet_basicblock if config.RESNET_MODE == 'preact' else resnet_basicblock
bottleneck = {
'resnet': resnet_bottleneck,
'preact': preresnet_bottleneck,
'se': se_resnet_bottleneck
}[config.RESNET_MODE]
num_blocks, block_func = {
18: ([2, 2, 2, 2], basicblock),
26: ([2, 2, 2, 2], bottleneck),
34: ([3, 4, 6, 3], basicblock),
50: ([3, 4, 6, 3], bottleneck),
101: ([3, 4, 23, 3], bottleneck),
152: ([3, 8, 36, 3], bottleneck)
}[depth]
logits = get_logit(image, num_blocks, block_func)
if is_training:
loss = cls_loss(logits, label)
#wd_cost = regularize_cost(
# '.*/W',
# l2_regularizer(1e-4), name='wd_cost')
self.cost = tf.add_n([loss], 'total_cost')
#add_moving_summary(self.cost)
else:
final_probs = tf.nn.sigmoid(logits, name="final_probs")
def _build_graph(self, inputs):
is_training = get_current_tower_context().is_training
image, label = inputs
tf.summary.image('viz', image, max_outputs=10)
image = image_preprocess(image, bgr=True)
image = tf.transpose(image, [0, 3, 1, 2])
conv4 = pretrained_resnet_conv4(image, config.RESNET_NUM_BLOCK[:3])
# B*c*h*w
conv5 = resnet_conv5(conv4, config.RESNET_NUM_BLOCK[-1])
# train head only
logits = cls_head('cls', conv5)
if is_training:
loss = cls_loss(logits, label)
wd_cost = regularize_cost('(?:group1|group2|group3|cls)/.*W',
l2_regularizer(1e-4),
name='wd_cost')
self.cost = tf.add_n([loss, wd_cost], 'total_cost')
add_moving_summary(self.cost, wd_cost)
else:
final_prob = tf.nn.sigmoid(logits)
tf.identity(final_prob, name='final_probs')
#final_label = tf.where(final_prob > 0.5, tf.ones_like(final_prob), tf.zeros_like(final_prob))
final_label = tf.round(final_prob)
tf.identity(final_label, name='final_labels')
def preprocess(self, image):
'''
norm图片之后,返回(N,C,H,W),caffe格式
:param image:
:return:
'''
image = tf.expand_dims(image, 0)
image = image_preprocess(image, bgr=True)
return tf.transpose(image, [0, 3, 1, 2])
def preprocess(self, image):
image = tf.expand_dims(image, 0)
image = image_preprocess(image, bgr=True)
return tf.transpose(image, [0, 3, 1, 2])
def _preprocess(self, image):
image = tf.expand_dims(image, 0)
image = image_preprocess(image, bgr=True)
return tf.transpose(image, [0, 3, 1, 2])
def _build_graph(self, inputs):
is_training = get_current_tower_context().is_training
image, anchor_labels, anchor_boxes, gt_boxes, gt_labels = inputs
image = tf.expand_dims(image, 0)
# FSxFSxNAx4 (FS=MAX_SIZE//ANCHOR_STRIDE)
with tf.name_scope('anchors'):
all_anchors = tf.constant(get_all_anchors(),
name='all_anchors',
dtype=tf.float32)
fm_anchors = tf.slice(
all_anchors, [0, 0, 0, 0],
tf.stack([
tf.shape(image)[1] // config.ANCHOR_STRIDE,
tf.shape(image)[2] // config.ANCHOR_STRIDE, -1, -1
]),
name='fm_anchors')
anchor_boxes_encoded = encode_bbox_target(anchor_boxes, fm_anchors)
image = image_preprocess(image, bgr=True)
image = tf.transpose(image, [0, 3, 1, 2])
# resnet50
featuremap = pretrained_resnet_conv4(image, [3, 4, 6])
rpn_label_logits, rpn_box_logits = rpn_head(featuremap)
rpn_label_loss, rpn_box_loss = rpn_losses(anchor_labels,
anchor_boxes_encoded,
rpn_label_logits,
rpn_box_logits)
decoded_boxes = decode_bbox_target(
rpn_box_logits, fm_anchors) # (fHxfWxNA)x4, floatbox
proposal_boxes, proposal_scores = generate_rpn_proposals(
decoded_boxes, tf.reshape(rpn_label_logits, [-1]),
tf.shape(image)[2:])
if is_training:
rcnn_sampled_boxes, rcnn_encoded_boxes, rcnn_labels = sample_fast_rcnn_targets(
proposal_boxes, gt_boxes, gt_labels)
boxes_on_featuremap = rcnn_sampled_boxes * (1.0 /
config.ANCHOR_STRIDE)
roi_resized = roi_align(featuremap, boxes_on_featuremap, 14)
feature_fastrcnn = resnet_conv5(roi_resized) # nxc
fastrcnn_label_logits, fastrcnn_box_logits = fastrcnn_head(
feature_fastrcnn, config.NUM_CLASS)
fastrcnn_label_loss, fastrcnn_box_loss = fastrcnn_losses(
rcnn_labels, rcnn_encoded_boxes, fastrcnn_label_logits,
fastrcnn_box_logits)
wd_cost = regularize_cost(
'(?:group1|group2|group3|rpn|fastrcnn)/.*W',
l2_regularizer(1e-4),
name='wd_cost')
self.cost = tf.add_n([
rpn_label_loss, rpn_box_loss, fastrcnn_label_loss,
fastrcnn_box_loss, wd_cost
], 'total_cost')
for k in self.cost, wd_cost:
add_moving_summary(k)
else:
roi_resized = roi_align(
featuremap, proposal_boxes * (1.0 / config.ANCHOR_STRIDE), 14)
feature_fastrcnn = resnet_conv5(roi_resized) # nxc
label_logits, fastrcnn_box_logits = fastrcnn_head(
feature_fastrcnn, config.NUM_CLASS)
label_probs = tf.nn.softmax(label_logits,
name='fastrcnn_all_probs') # NP,
labels = tf.argmax(label_logits, axis=1)
fg_ind, fg_box_logits = fastrcnn_predict_boxes(
labels, fastrcnn_box_logits)
fg_label_probs = tf.gather(label_probs,
fg_ind,
name='fastrcnn_fg_probs')
fg_boxes = tf.gather(proposal_boxes, fg_ind)
fg_box_logits = fg_box_logits / tf.constant(
config.FASTRCNN_BBOX_REG_WEIGHTS)
decoded_boxes = decode_bbox_target(fg_box_logits,
fg_boxes) # Nfx4, floatbox
decoded_boxes = tf.identity(decoded_boxes,
name='fastrcnn_fg_boxes')
def _preprocess(self, image):
image = tf.expand_dims(image, 0)
image = image_preprocess(image, bgr=True)
return image
