Overload this function in your dataset. This should return number of sequences in the dataset.

Returns size of the dataset returns: int - number of samples in the dataset

perform evaluation and save ckpt.

perform evaluation and save ckpt.

Parse bbox and mask annotation. Args: ann_info (list[dict]): Annotation info of an image. with_mask (bool): Whether t

Accumulate results in a single video.

Computes the precision@k for the specified values of k

inp_lbs: PyTorch tensor (F, K, H, W) --> (F, K, H*s, W*s) dh: destination h, dw: destination w

inp_lbs: PyTorch tensor (F, K, H, W) --> (F, K, H*s, W*s) dh: destination h, dw: destination w

Run all_gather on arbitrary picklable data (not necessarily tensors) Args: data: any picklable object Returns: list[data]

Args: pred: Nx4 predicted bounding boxes target: Nx4 target bounding boxes Both are in the form of FCOS prediction (l, t,

Concatenates two dicts without copying internal data.

Cosine learning rate