a677be4ce43130c024c8e0a48738fdb8d66b70ca,keras_retinanet/preprocessing/coco.py,CocoIterator,next,#CocoIterator#,59

Before Change

        assert(batch_size == 1), "Currently only batch_size=1 is allowed."

        boxes_batch = np.zeros((batch_size, 0, 5), dtype=keras.backend.floatx())

        for batch_index, image_index in enumerate(selection):
            coco_image = self.coco.loadImgs(self.image_ids[image_index])[0]
            path  = os.path.join(self.data_dir, self.set_name, coco_image["file_name"])
            image = cv2.imread(path, cv2.IMREAD_COLOR)
            image, image_scale = resize_image(image, min_side=self.image_min_side, max_side=self.image_max_side)

            // copy image to image batch (currently only batch_size==1 is allowed)
            image_batch = np.expand_dims(image, axis=0).astype(keras.backend.floatx())

            // set ground truth boxes
            annotations_ids = self.coco.getAnnIds(imgIds=coco_image["id"], iscrowd=False)

            // some images appear to miss annotations (like image with id 257034)
            if len(annotations_ids) == 0:
                return self.next()

            // parse annotations
            annotations = self.coco.loadAnns(annotations_ids)
            for idx, a in enumerate(annotations):
                box = np.zeros((1, 1, 5), dtype=keras.backend.floatx())
                box[0, 0, :4] = a["bbox"]
                box[0, 0, 4]  = a["category_id"]
                boxes_batch = np.append(boxes_batch, box, axis=1)

            // transform from [x, y, w, h] to [x1, y1, x2, y2]
            boxes_batch[batch_index, :, 2] = boxes_batch[batch_index, :, 0] + boxes_batch[batch_index, :, 2]
            boxes_batch[batch_index, :, 3] = boxes_batch[batch_index, :, 1] + boxes_batch[batch_index, :, 3]

            // scale the ground truth boxes to the selected image scale
            boxes_batch[batch_index, :, :4] *= image_scale

        // randomly transform images and boxes simultaneously
        image_batch, boxes_batch = random_transform_batch(image_batch, boxes_batch, self.image_data_generator)

        // convert the image to zero-mean
        image_batch = keras.applications.imagenet_utils.preprocess_input(image_batch)
        image_batch = self.image_data_generator.standardize(image_batch)

        return [image_batch, boxes_batch], None

After Change

            // generate the label and regression targets
            labels, reg_targets = anchor_targets(image, boxes_batch[0])
            target = np.append(reg_targets, np.expand_dims(labels, axis=1), axis=1)

            // convert target to batch (currently only batch_size = 1 is allowed)
            target_batch = np.expand_dims(target, axis=0)

        // convert the image to zero-mean
        image_batch = keras.applications.imagenet_utils.preprocess_input(image_batch)
        image_batch = self.image_data_generator.standardize(image_batch)

        return image_batch, target_batch

Instances