Compute the d_change metric proposed in Box2Pix: https://lmb.informatik.uni-freiburg.de/Publications/2018/UB18/paper-box2pix.pdf Inp

coeffs should be size [num_pos, num_coeffs] instance_t should be size [num_pos] and be values from 0 to num_instances-1

Crops the gt masks using the predicted bboxes, scales them down, and outputs the BCE loss.

Dumps it in the format for my web app. Warning: bad code ahead!

Encode bboxes matched with each prior into the format produced by the network. See decode for more details on this format. Note that enco

Focal loss as described in https://arxiv.org/pdf/1708.02002.pdf Adapted from https://github.com/clcarwin/focal_loss_pytorch/blob/mast

Instead of using softmax, use class[0] to be the objectness score and do sigmoid focal loss on that. Then for the rest of the classes

Focal loss but using sigmoid like the original paper. Note: To make things mesh easier, the network still predicts 81 class confidenc

Warning: result not cached.

transformed_cat_id is [0,80) as indices in cfg.dataset.class_names

Initializes the backbone weights for training.

Initialize the console to be able to use ANSI escape characters on Windows.

We resize both tensors to [A,B,2] without new malloc: [A,2] -> [A,1,2] -> [A,B,2] [B,2] -> [1,B,2] -> [A,B,2] Then we compute the area of

We resize both tensors to [A,B,2] without new malloc: [A,2] -> [A,1,2] -> [A,B,2] [B,2] -> [1,B,2] -> [A,B,2] Then we compute the area of

We resize both tensors to [A,B,2] without new malloc: [A,2] -> [A,1,2] -> [A,B,2] [B,2] -> [1,B,2] -> [A,B,2] Then we compute the area of

Compute the jaccard overlap of two sets of boxes. The jaccard overlap is simply the intersection over union of two boxes. Here we operate on

Compute the jaccard overlap of two sets of boxes. The jaccard overlap is simply the intersection over union of two boxes. Here we operate on

Compute the jaccard overlap of two sets of boxes. The jaccard overlap is simply the intersection over union of two boxes. E.g.: A ∩ B

Compute the jaccard overlap of two sets of boxes. The jaccard overlap is simply the intersection over union of two boxes. E.g.: A ∩ B

Utility function for computing log_sum_exp while determining This will be used to determine unaveraged confidence loss across all examples in

Inputs inputs are matricies of size _ x N. Output is size _1 x _2. Note: if iscrowd is True, then mask2 should be the crowd.

Match each prior box with the ground truth box of the highest jaccard overlap, encode the bounding boxes, then return the matched indices corr

Note: If undo_transform=False then im_h and im_w are allowed to be None.

Converts the Backbone to a TRTModule.

Converts FlowNet to a TRTModule.

Converts FPN to a TRTModule.

Converts Prediction Head to a TRTModule.

Converts ProtoNet to a TRTModule.

Converts SPA to a TRTModule.

Returns the total amount accumulated across all functions in seconds.

Transform e.g. features.24.bias to layers.4.1.bias