def _gen_anchors(self, config, feature_map_shape):
model = FasterRCNN(config)
results = model._generate_anchors(feature_map_shape)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
results = sess.run(results)
return results
def _gen_anchors(self, config, feature_map_shape):
model = FasterRCNN(config)
results = model._generate_anchors(feature_map_shape)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
results = sess.run(results)
return results
def _get_losses(self, config, prediction_dict, image_size):
image = tf.placeholder(tf.float32, shape=self.image.shape)
gt_boxes = tf.placeholder(tf.float32, shape=self.gt_boxes.shape)
model = FasterRCNN(config)
model(image, gt_boxes)
all_losses = model.loss(prediction_dict, return_all=True)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
all_losses = sess.run(all_losses)
return all_losses
def _get_losses(self, config, prediction_dict, image_size):
image = tf.placeholder(
tf.float32, shape=self.image.shape)
gt_boxes = tf.placeholder(
tf.float32, shape=self.gt_boxes.shape)
model = FasterRCNN(config)
model(image, gt_boxes, is_training=True)
all_losses = model.loss(prediction_dict, return_all=True)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
all_losses = sess.run(all_losses)
return all_losses
def _run_network(self):
image = tf.placeholder(tf.float32, shape=self.image.shape)
gt_boxes = tf.placeholder(tf.float32, shape=self.gt_boxes.shape)
model = FasterRCNN(self.config)
results = model(image, gt_boxes)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
results = sess.run(results,
feed_dict={
gt_boxes: self.gt_boxes,
image: self.image,
})
return results