Collected results in distributed environments. Args: target (str): Target results to collect. Returns: Optio

Returns a random color

Parse channel order of inputs in batch.

Parse channel index of inputs.

Update `padding_info` and `channel_order` to metainfo of. *a batch of `data_samples`*. For channel order, we consider same field amon

img: input image mask: input mask n: mosaic size model : squa_avg squa_mid squa_random squa_avg_circle_edge rect_avg rect_ratio:

1x1 convolution

3x3 convolution with padding

Destruct padding, normalization and convert channel order to BGR if could. If `data_samples` is a list, outputs will be destructed as a

Forward tensor. Returns result of simple forward. Args: inputs (torch.Tensor): batch input tensor collated by :at

Trans image to tensor. Args: img (np.ndarray): The original image. Returns: Tensor: The output tensor.

Generate the mesh grid, centering at zero. Args: kernel_size (int): Returns: xy (ndarray): with the shape (kernel_size, kernel

Runs Non-Maximum Suppression (NMS) on inference results Returns: list of detections, on (n,6) tensor per image [xyxy, conf, cls]

Obtain data of key from data_sample and converse data to device. Args: data_sample (dict): A dict of data sample. key (str): The k

Calculate the rotated sigma matrix (two dimensional matrix). Args: sig_x (float): sig_y (float): theta (float): Radian mea

Convert (x1, y1, x2, y2) to (x, y, w, h) with x, y the center of the bbox and w, h the width and height

Returns a random alpha value for the watermark Args: position (dict): The position of the watermark rotation (int): The rota

Returns a rotation angle from a position TODO: TO FIX

Calculate Gram matrix. Args: x (torch.Tensor): Tensor with shape of (n, c, h, w). Returns: torch.Tensor: Gra

Preprocess dict type inputs. Args: batch_inputs (dict): Input dict. data_samples (List[DataSample], optional): The da

Preprocess a list of image tensor and update metainfo to corresponding data samples. Args: tensor_list (List[Tensor]): Im

Preprocess a batch of image tensor and update metainfo to corresponding data samples. Args: inputs (Tensor): Image tensor

Processes the buffer to encode frames.

returns Scene if it fits the criteria for a valid completed scene like min/max length

2D sinc filter, ref: https://dsp.stackexchange.com/questions/58301/2-d-circularly-symmetric-low-pass-filter Args: cutoff (float): cutoff f