Add more information about dataset

[IcyFeather233]

In testenv.yaml configuration, train_url and test_url is configured.

But no document is about how these two data files should be prepared.

For example, what data format is needed? I think if there is a example, it will be better for people new to this project to use.

The document should cover TXT, CSV, JSON, which is in load data class

[IcyFeather233]

By the way, I think the JSONDataParse class is not standard, it seems that it's only used for coco, which is a image dataset, I think it should be a universal json data parse class:

class JSONDataParse(BaseDataSource, ABC):
    """
    json file which contain Structured Data parser
    """

    def __init__(self, data_type, func=None):
        super(JSONDataParse, self).__init__(data_type=data_type, func=func)

    def parse(self, *args, **kwargs):
        DIRECTORY = "train"
        LABEL_PATH = "*/gt/gt_val_half.txt"
        filepath = Path(*args)
        self.data_dir = Path(Path(filepath).parents[1], DIRECTORY)
        self.coco = COCO(filepath)
        self.ids = self.coco.getImgIds()
        self.class_ids = sorted(self.coco.getCatIds())
        self.annotations = [self.load_anno_from_ids(_ids) for _ids in self.ids]
        self.x = {"data_dir": self.data_dir, "coco": self.coco, "ids": self.ids, "class_ids": self.class_ids, "annotations": self.annotations}
        self.y = [f for f in self.data_dir.glob(LABEL_PATH)]

    def load_anno_from_ids(self, id_):
        im_ann = self.coco.loadImgs(id_)[0]
        width = im_ann["width"]
        height = im_ann["height"]
        frame_id = im_ann["frame_id"]
        video_id = im_ann["video_id"]
        anno_ids = self.coco.getAnnIds(imgIds=[int(id_)], iscrowd=False)
        annotations = self.coco.loadAnns(anno_ids)
        objs = []
        for obj in annotations:
            x1 = obj["bbox"][0]
            y1 = obj["bbox"][1]
            x2 = x1 + obj["bbox"][2]
            y2 = y1 + obj["bbox"][3]
            if obj["area"] > 0 and x2 >= x1 and y2 >= y1:
                obj["clean_bbox"] = [x1, y1, x2, y2]
                objs.append(obj)

        num_objs = len(objs)
        res = np.zeros((num_objs, 6))

        for ix, obj in enumerate(objs):
            cls = self.class_ids.index(obj["category_id"])
            res[ix, 0:4] = obj["clean_bbox"]
            res[ix, 4] = cls
            res[ix, 5] = obj["track_id"]

        file_name = im_ann["file_name"] if "file_name" in im_ann else "{:012}".format(id_) + ".jpg"
        img_info = (height, width, frame_id, video_id, file_name)

        del im_ann, annotations

        return (res, img_info, file_name)