my_datasets/refexp2seq.py

# Copyright (c) Aishwarya Kamath & Nicolas Carion. Licensed under the Apache License 2.0. All Rights Reserved
# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved
# For building refcoco, refcoco+, refcocog datasets
"""
COCO dataset which returns image_id for evaluation.
Mostly copy-paste from https://github.com/pytorch/vision/blob/13b35ff/references/detection/coco_utils.py
"""
from pathlib import Path

import torch
import torch.utils.data
import torchvision
from pycocotools import mask as coco_mask

import random
import numpy as np
from PIL import Image

import datasets.transforms_video as T
from datasets.image_to_seq_augmenter import ImageToSeqAugmenter

from util.box_ops import masks_to_boxes


class ModulatedDetection(torchvision.datasets.CocoDetection):
    def __init__(self, img_folder, ann_file, num_frames, transforms, return_masks):
        super(ModulatedDetection, self).__init__(img_folder, ann_file)
        self._transforms = transforms
        self.prepare = ConvertCocoPolysToMask(return_masks)
        self.num_frames = num_frames
        self.augmenter = ImageToSeqAugmenter(perspective=True, affine=True, motion_blur=True,
                                             rotation_range=(-20, 20), perspective_magnitude=0.08,
                                             hue_saturation_range=(-5, 5), brightness_range=(-40, 40),
                                             motion_blur_prob=0.25, motion_blur_kernel_sizes=(9, 11),
                                             translate_range=(-0.1, 0.1))

    def apply_random_sequence_shuffle(self, images, instance_masks):
        perm = list(range(self.num_frames))
        random.shuffle(perm)
        images = [images[i] for i in perm]
        instance_masks = [instance_masks[i] for i in perm]
        return images, instance_masks

    def __getitem__(self, idx):
        instance_check = False
        while not instance_check:
            img, target = super(ModulatedDetection, self).__getitem__(idx)
            image_id = self.ids[idx]
            coco_img = self.coco.loadImgs(image_id)[0]
            caption = coco_img["caption"]
            dataset_name = coco_img["dataset_name"] if "dataset_name" in coco_img else None
            target = {"image_id": image_id, "annotations": target, "caption": caption}
            img, target = self.prepare(img, target)

            # for a image, we rotate it to form a clip
            seq_images, seq_instance_masks = [img], [target['masks'].numpy()]
            numpy_masks = target['masks'].numpy() # [1, H, W]

            numinst = len(numpy_masks)
            assert numinst == 1
            for t in range(self.num_frames - 1):
                im_trafo, instance_masks_trafo = self.augmenter(np.asarray(img), numpy_masks)
                im_trafo = Image.fromarray(np.uint8(im_trafo))
                seq_images.append(im_trafo)
                seq_instance_masks.append(np.stack(instance_masks_trafo, axis=0))
            seq_images, seq_instance_masks = self.apply_random_sequence_shuffle(seq_images, seq_instance_masks)
            output_inst_masks = []
            for inst_i  in range(numinst):
                inst_i_mask = []
                for f_i in range(self.num_frames):
                    inst_i_mask.append(seq_instance_masks[f_i][inst_i])
                output_inst_masks.append( np.stack(inst_i_mask, axis=0) )
            
            output_inst_masks = torch.from_numpy( np.stack(output_inst_masks, axis=0) )         
            target['masks'] = output_inst_masks.flatten(0,1)            # [t, h, w]
            target['boxes'] = masks_to_boxes(target['masks'])           # [t, 4]
            target['labels'] = target['labels'].repeat(self.num_frames) # [t,]

            if self._transforms is not None:
                img, target = self._transforms(seq_images, target)
            target["dataset_name"] = dataset_name
            for extra_key in ["sentence_id", "original_img_id", "original_id", "task_id"]:
                if extra_key in coco_img:
                    target[extra_key] = coco_img[extra_key] # box xyxy -> cxcywh
            # FIXME: handle "valid", since some box may be removed due to random crop
            if torch.any(target['valid'] == 1):  # at leatst one instance
                instance_check = True
            else:
                idx = random.randint(0, self.__len__() - 1)

        # set the gt box of empty mask to [0, 0, 0, 0]
        for inst_id in range(len(target['boxes'])):
            if target['masks'][inst_id].max()<1:
                target['boxes'][inst_id] =  torch.zeros(4).to(target['boxes'][inst_id]) 

        target['boxes']=target['boxes'].clamp(1e-6)
        return torch.stack(img,dim=0), target


def convert_coco_poly_to_mask(segmentations, height, width):
    masks = []
    for polygons in segmentations:
        rles = coco_mask.frPyObjects(polygons, height, width)
        mask = coco_mask.decode(rles)
        if len(mask.shape) < 3:
            mask = mask[..., None]
        mask = torch.as_tensor(mask, dtype=torch.uint8)
        mask = mask.any(dim=2)
        masks.append(mask)
    if masks:
        masks = torch.stack(masks, dim=0)
    else:
        masks = torch.zeros((0, height, width), dtype=torch.uint8)
    return masks


class ConvertCocoPolysToMask(object):
    def __init__(self, return_masks=False):
        self.return_masks = return_masks

    def __call__(self, image, target):
        w, h = image.size

        image_id = target["image_id"]
        image_id = torch.tensor([image_id])

        anno = target["annotations"]
        caption = target["caption"] if "caption" in target else None

        anno = [obj for obj in anno if "iscrowd" not in obj or obj["iscrowd"] == 0]

        boxes = [obj["bbox"] for obj in anno]
        # guard against no boxes via resizing
        boxes = torch.as_tensor(boxes, dtype=torch.float32).reshape(-1, 4)
        boxes[:, 2:] += boxes[:, :2] # xminyminwh -> xyxy
        boxes[:, 0::2].clamp_(min=0, max=w)
        boxes[:, 1::2].clamp_(min=0, max=h)

        classes = [obj["category_id"] for obj in anno]
        classes = torch.tensor(classes, dtype=torch.int64)

        if self.return_masks:
            segmentations = [obj["segmentation"] for obj in anno]
            masks = convert_coco_poly_to_mask(segmentations, h, w)

        # keep the valid boxes
        keep = (boxes[:, 3] > boxes[:, 1]) & (boxes[:, 2] > boxes[:, 0])
        boxes = boxes[keep]
        classes = classes[keep]
        if self.return_masks:
            masks = masks[keep]

        target = {}
        target["boxes"] = boxes
        target["labels"] = classes
        if caption is not None:
            target["caption"] = caption
        if self.return_masks:
            target["masks"] = masks
        target["image_id"] = image_id

        # for conversion to coco api
        area = torch.tensor([obj["area"] for obj in anno])
        iscrowd = torch.tensor([obj["iscrowd"] if "iscrowd" in obj else 0 for obj in anno])
        target["area"] = area[keep]
        target["iscrowd"] = iscrowd[keep]
        target["valid"] = torch.tensor([1])
        target["orig_size"] = torch.as_tensor([int(h), int(w)])
        target["size"] = torch.as_tensor([int(h), int(w)])
        return image, target


def make_coco_transforms(image_set, max_size):
    normalize = T.Compose([
        T.ToTensor(),
        T.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
    ])

    scales = [288, 320, 352, 392, 416, 448, 480, 512]

    if image_set == 'train':
        return T.Compose([
            T.RandomHorizontalFlip(),
            T.PhotometricDistort(),
            T.RandomSelect(
                T.Compose([
                    T.RandomResize(scales, max_size=max_size),
                    T.Check(),
                ]),
                T.Compose([
                    T.RandomResize([400, 500, 600]),
                    T.RandomSizeCrop(384, 600),
                    T.RandomResize(scales, max_size=max_size),
                    T.Check(),
                ])
            ),
            normalize,
        ])

    if image_set == "val":
        return T.Compose(
            [
                T.RandomResize([360], max_size=640),
                normalize,
            ]
        )

    raise ValueError(f"unknown {image_set}")


def build(dataset_file, image_set, args):
    root = Path(args.coco_path)
    assert root.exists(), f"provided COCO path {root} does not exist"
    mode = "instances"
    dataset = dataset_file
    PATHS = {
        "train": (root / "train2014", root / dataset / f"{mode}_{dataset}_train.json"),
        "val": (root / "train2014", root / dataset / f"{mode}_{dataset}_val.json"),
    }

    img_folder, ann_file = PATHS[image_set]
    dataset = ModulatedDetection(
        img_folder,
        ann_file,
        num_frames=args.num_frames,
        transforms=make_coco_transforms(image_set, args.max_size),
        return_masks=args.masks,
    )
    return dataset