def draw_predictions(dataset, pred_bboxes, pred_labels, pred_scores, labels,
batch_size):
for batch_id, image_data in enumerate(dataset):
imgs, _, _ = image_data
img_size = imgs.shape[1]
start = batch_id * batch_size
end = start + batch_size
batch_bboxes, batch_labels, batch_scores = pred_bboxes[
start:end], pred_labels[start:end], pred_scores[start:end]
for i, img in enumerate(imgs):
denormalized_bboxes = bbox_utils.denormalize_bboxes(
batch_bboxes[i], img_size, img_size)
draw_bboxes_with_labels(img, denormalized_bboxes, batch_labels[i],
batch_scores[i], labels)
def patch(img, gt_boxes):
"""Generating random patch and adjusting image and ground truth objects to this patch.
After this operation some of the ground truth boxes / objects could be removed from the image.
However, these objects are not excluded from the output, only the coordinates are changed as zero.
inputs:
img = (height, width, depth)
gt_boxes = (ground_truth_object_count, [y1, x1, y2, x2])
in normalized form [0, 1]
outputs:
modified_img = (final_height, final_width, depth)
modified_gt_boxes = (ground_truth_object_count, [y1, x1, y2, x2])
in normalized form [0, 1]
"""
img_shape = tf.shape(img)
height, width = tf.cast(img_shape[0],
tf.float32), tf.cast(img_shape[1], tf.float32)
# Denormalizing bounding boxes for further operations
denormalized_gt_boxes = bbox_utils.denormalize_bboxes(
gt_boxes, height, width)
# Randomly expand image and adjust bounding boxes
img, denormalized_gt_boxes, height, width = randomly_apply_operation(
expand_image, img, denormalized_gt_boxes, height, width)
# Generate random patches
patches = generate_random_patches(height, width)
# Calculate jaccard/iou value for each bounding box
iou_map = bbox_utils.generate_iou_map(patches,
denormalized_gt_boxes,
transpose_perm=[1, 0])
# Check each ground truth box center in the generated patch and return a boolean condition list
center_in_patch_condition = get_center_in_patch_condition(
patches, denormalized_gt_boxes)
# Get random minimum overlap value
min_overlap = get_random_min_overlap()
# Check and merge center condition and minimum intersection condition
valid_patch_condition = tf.logical_and(center_in_patch_condition,
tf.greater(iou_map, min_overlap))
# Check at least one valid patch then apply patch
img, denormalized_gt_boxes, height, width = tf.cond(
tf.reduce_any(valid_patch_condition), lambda: select_and_apply_patch(
img, denormalized_gt_boxes, patches, valid_patch_condition,
center_in_patch_condition), lambda:
(img, denormalized_gt_boxes, height, width))
# Finally normalized ground truth boxes
gt_boxes = bbox_utils.normalize_bboxes(denormalized_gt_boxes, height,
width)
gt_boxes = tf.clip_by_value(gt_boxes, 0, 1)
#
return img, gt_boxes
total_landmarks = hyper_params["total_landmarks"]
landmark_variances = total_landmarks * variances[0:2]
variances += landmark_variances
for image_data in test_data:
img, _, _ = image_data
pred_deltas, pred_scores = model.predict_on_batch(img)
pred_deltas *= variances
#
pred_bboxes_and_landmarks = bbox_utils.get_bboxes_and_landmarks_from_deltas(
prior_boxes, pred_deltas)
pred_bboxes_and_landmarks = tf.clip_by_value(pred_bboxes_and_landmarks, 0,
1)
#
pred_scores = tf.cast(pred_scores, tf.float32)
#
weighted_suppressed_data = bbox_utils.weighted_suppression(
pred_scores[0], pred_bboxes_and_landmarks[0])
#
weighted_bboxes = weighted_suppressed_data[..., 0:4]
weighted_landmarks = weighted_suppressed_data[..., 4:]
#
denormalized_bboxes = bbox_utils.denormalize_bboxes(
weighted_bboxes, img_size, img_size)
weighted_landmarks = tf.reshape(weighted_landmarks,
(-1, total_landmarks, 2))
denormalized_landmarks = landmark_utils.denormalize_landmarks(
weighted_landmarks, img_size, img_size)
drawing_utils.draw_bboxes_with_landmarks(img[0], denormalized_bboxes,
denormalized_landmarks)
for image_data in test_data:
img, _, _ = image_data
pred_bbox_deltas, pred_labels = ssd_model.predict_on_batch(img)
#
pred_bbox_deltas *= hyper_params["variances"]
pred_bboxes = bbox_utils.get_bboxes_from_deltas(prior_boxes,
pred_bbox_deltas)
pred_bboxes = tf.clip_by_value(pred_bboxes, 0, 1)
#
pred_labels = tf.cast(pred_labels, tf.float32)
reshaped_pred_bboxes = tf.reshape(
pred_bboxes, (batch_size, pred_bbox_deltas.shape[1], 1, 4))
# Remove background predictions
pred_labels_map = tf.argmax(pred_labels, 2, output_type=tf.int32)
valid_cond = tf.not_equal(pred_labels_map, bg_id)
#
valid_bboxes = tf.expand_dims(reshaped_pred_bboxes[valid_cond], 0)
valid_labels = tf.expand_dims(pred_labels[valid_cond], 0)
#
nms_bboxes, nmsed_scores, nmsed_classes, valid_detections = bbox_utils.non_max_suppression(
valid_bboxes,
valid_labels,
max_output_size_per_class=10,
max_total_size=200,
score_threshold=0.5)
denormalized_bboxes = bbox_utils.denormalize_bboxes(
nms_bboxes[0], img_size, img_size)
drawing_utils.draw_bboxes_with_labels(img[0], denormalized_bboxes,
nmsed_classes[0], nmsed_scores[0],
labels)