Dynamically creates a new type that inherits the type of a given instance and overrides the `step` method to add gradient clipping

Gets the batch size of the output given the batch size of the input prompts.

Get a dummy object pointer based on an empty mask on the current frame.

Compute stability scores of the mask logits based on the IoU between upper and lower thresholds.

Args: input (dict): one dataset dict with "image" field being a CHW tensor Returns: dict: one output dict

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

Fuse the current frame's visual feature map with previous memory.

Reset all tracking inputs and results across the videos.

Run the memory encoder on `high_res_masks`. This is usually after applying non-overlapping constraints to object scores. Since their

Run the memory encoder on `high_res_masks`. This is usually after applying non-overlapping constraints to object scores. Since their

Directly turn binary `mask_inputs` into a output mask logits without using SAM. (same input and output shapes as in _forward_sam_head

Whether to use multimask output in the SAM head.

Add new points to a frame.

Returns the "name" of an annotation, optionally including the attributes. :param attributes: The attribute dictionary. :param category_na

Generates point grids for all crop layers.

Generates a 2D grid of points evenly spaced in [0,1]x[0,1].

We estimate uncerainty as L1 distance between 0.0 and the logit prediction in 'logits' for the foreground class in `classes`. Args:

We estimate uncerainty as L1 distance between 0.0 and the logit prediction in 'logits' for the foreground class in `classes`. Args:

Remove all input points or mask in a specific frame for a given object.

Remove all input points or mask in a specific frame for a given object.

Disable the redundant torch default initialization to accelerate model creation.

从原始文本中提取intera任务的坐标（框/点） 返回：coord_tensor（框：[1,4]；点：[n,2]，n=1-3；无坐标：None）

Identical to the corresponding method in the parent (SAM2VideoPredictor), but cloning the backbone features and pos encoding to enabl

Generates a list of crop boxes of different sizes. Each layer has (2**i)**2 boxes for the ith layer.

Get 1D sine positional embedding as in the original Transformer paper.

Return an activation function given a string