Given a visual crop and frames from a clip, retrieve the bounding box proposal from each frame that is most similar to the visual crop.

Process the question to make it canonical.

Convert the format from JSON files. fps, num_frames, timestamps, sentences, exact_times, annotation_uids, query_idx.

Remove an instance from the register. Args: instance_id (str): the instance id (number) to remove Returns:

Removes keypoints too close to the border

init_state: initial state of the tracked obj in the frame (gathered from the detection stage) [x,y,sx, sy] (relat

init_state: initial state of the tracked obj in the frame (gathered from the detection stage) [x,y,sx, sy] (relat

Interpolate descriptors at keypoint locations

Saves a checkpoint of the network and other variables.

Compute the temporal intersection over union between a target segment and all the test segments. Parameters ---------- target_segment

Produce additional matches for predictions that have only low-quality matches. Specifically, for each ground-truth find the set of pr

Create configs and perform basic setups.

Fast Non-maximum suppression to remove nearby points

Computes intersection between two bounding boxes

Computes iou between two bounding boxes

For each bounding box in gt, find a match in pred and measure the per-frame IoU. Set IoU to zero if no match is found. Note: This assume

For each BBox in each candidate response track (ground-truth), find the IoU b/w itself and a BBox from the target response track (prediction)

Computes tube-iou between two response track windows. Note: This assumes that each bbox in the list corresponds to a different frame. Can

Must be deterministic

Transform the box co-ordinates from the original image co-ordinates to the co-ordinates of the cropped image args: box_in - the box for wh

Visualizes an annotation from the visual-queries task

Computes average precision for each class in the subset.

Computes average precision for each class in the subset.

Find the points that will be newly added Parameters ---------- X : ndarray of shape (F, 3) 3D points track_i : ndarray o

Run bundle adjustment Parameters ---------- P : ndarray of shape (K, 3, 4) Set of reconstructed camera poses X : ndarray

Refine the triangulated points Parameters ---------- X : ndarray of shape (n, 3) 3D points P1 : ndarray of shape (3, 4)