Concatenates a list of Boxes into a single Boxes Arguments: boxes_list (list[Boxes]) Returns: Boxes

Args: box: can be a k-tuple, k-list or an Nxk array/tensor, where k = 4 or 5 from_mode, to_mode (BoxMode) Re

Clip (in place) the boxes by limiting x coordinates to the range [0, width] and y coordinates to the range [0, height]. Args

Convert a rgb color to hex color. Args: rgb_color (tuple/list of ints): RGB color in tuple or list format. Returns: str

Read and unpack the next bytes from a binary file. :param fid: :param num_bytes: Sum of combination of {2, 4, 8}, e.g. 2, 6, 16, 30, etc.

Clone the Boxes. Returns: Boxes

Clone a tensor if and only if `self.compile_all_components` is True.

Concatenates two right-padded sequences, such that the resulting sequence is contiguous and also right-padded. Following pytorch's conve

Copy data to the specified device

构建 stem -> instances 映射，过滤 ignore_labels。

Wraps a given module to enable or disable activation checkpointing. Activation checkpointing (gradient checkpointing) trades compute for mem

Render the scene. Background tensor (bg_color) must be on GPU!

Sample k index pairs with covisibility above covis_threshold and both valid, without repeating images. Optional motion filters (if frames provide

Prepare and flatten visual features (same as in MDETR_API model).

Create a Color instance from a hex string. Args: color_hex (str): Hex string of the color. Returns:

Args: boxes (Boxes, RotatedBoxes or ndarray): either a :class:`Boxes`, or an Nx4 numpy array of XYXY_ABS format f

计算3D bbox的中心点（世界坐标）。

检查两个3D bbox是否有交集或距离过近。 Args: bbox1: 第一个bbox（8个3D点） bbox2: 第二个bbox（8个3D点） min_separation: 最小分离距离（米），如果两个bbox的最近距离小于此值，

通过 mask_path 从 mask_index.json 中查找对应的 mask RLE 数据。

fallback: 当 mask_index.json 里缺失 mask_rle 时，从 bbox json 的 instance.mask_encodings 获取。 对齐规则：优先 images 与 mask_encodings 同长度且按 index 对齐；否则尝试 base

从 RLE mask 计算 2D 边界框 (x_min, y_min, x_max, y_max)。

从 RLE mask 计算 2D 中心点。

Robustly project a near-rotation matrix to SO(3) using SVD. Ensures det(R)=+1.

Computes the area of all the boxes. Returns: torch.Tensor: a vector with areas of each box.

解码RLE格式的mask

Args: filepath (str): a string that contains the absolute path, including the file name, where the visualized ima

Apply non-overlapping constraints to the object scores in pred_masks. Here we keep only the highest scoring object at each spatial lo

Depending on the brightness_factor, gives a lighter or darker color i.e. a color with less or more saturation than the original color

Consolidate the per-object temporary outputs in `temp_output_dict_per_obj` on a frame into a single output for all objects, including

Get the total number of unique object ids received so far in this session.

Resize the object scores to the original video resolution (video_res_masks) and apply non-overlapping constraints for final output.

Initialize an inference state from `resource_path` (an image or a video).

Prepare inference_state and consolidate temporary outputs before tracking.

Overlays masklets and bounding boxes on a single image frame. Args: img: np.ndarray, shape (H, W, 3), uint8 or float32 in [0,255] or

Split a multi-object output into per-object output slices and add them into `output_dict_per_obj`. The resulting slices share the sam

通过 mask_path 从 mask_index.json 中查找对应的 mask RLE 数据。 Args: mask_index: mask_index.json 的内容 mask_path: mask 路径（来自 instance 的 ima

fallback: 当 mask_index.json 里缺失 mask_rle 时，从 bbox json 的 instance.mask_encodings 获取。 对齐规则：优先 images 与 mask_encodings 同长度且按 index 对齐；否则尝试 base

从 RLE mask 计算 2D 边界框 (x_min, y_min, x_max, y_max)。 Args: mask_rle: RLE 格式的 mask 字典，包含 'size' 和 'counts' Returns: 2D

从 RLE mask 计算 2D 中心点。 Args: mask_rle: RLE 格式的 mask 字典，包含 'size' 和 'counts' Returns: 2D 中心点 (x, y)，如果解码失败则返回 None

Locate GPU ID for a given object.

Get a statistics string for live sessions and their GPU usage.

Map client-side object id to model-side object index.

action_history is used to automatically decide what to do during propagation. action_type: one of ["add", "remove", "refine"] + ["pro

Clone the CUDA output tensors of a function to avoid in-place operations. This wrapper is useful when working with torch.compile to prevent

Computes connected components labeling on a batch of 2D tensors, using the best available backend. Args: input_tensor (torch.Tensor)

Args: x_data (list[int]): a list containing x values of all the points being drawn. Length of list should match t

Args: text (str): class label position (tuple): a tuple of the x and y coordinates to place text on image.

A post processor to fill small holes in mask scores with area under `max_area`. Holes are those small connected components in either backgrou

A command line logger.

Compute the IoU (Intersection over Union) between predicted masks and ground truth masks. Args: - pred_masks: (N, H, W) bool Tensor, co