Spaces:
Sleeping
Sleeping
| # Copyright (c) Aishwarya Kamath & Nicolas Carion. Licensed under the Apache License 2.0. All Rights Reserved | |
| # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved | |
| """ | |
| Modules to compute the matching cost and solve the corresponding LSAP. | |
| """ | |
| import torch | |
| from scipy.optimize import linear_sum_assignment | |
| from torch import nn | |
| import pdb | |
| from maskrcnn_benchmark.layers.set_loss import generalized_box_iou, box_iou | |
| class HungarianMatcherCustom(nn.Module): | |
| """This class computes an assignment between the targets and the predictions of the network | |
| For efficiency reasons, the targets don't include the no_object. Because of this, in general, | |
| there are more predictions than targets. In this case, we do a 1-to-1 matching of the best predictions, | |
| while the others are un-matched (and thus treated as non-objects). | |
| """ | |
| def __init__(self, cost_class: float = 1, cost_bbox: float = 1, cost_giou: float = 1, special = False): | |
| """Creates the matcher | |
| Params: | |
| cost_class: This is the relative weight of the classification error in the matching cost | |
| cost_bbox: This is the relative weight of the L1 error of the bounding box coordinates in the matching cost | |
| cost_giou: This is the relative weight of the giou loss of the bounding box in the matching cost | |
| """ | |
| super().__init__() | |
| self.cost_class = cost_class | |
| self.cost_bbox = cost_bbox | |
| self.cost_giou = cost_giou | |
| self.norm = nn.Softmax(-1) | |
| self.special = special | |
| assert cost_class != 0 or cost_bbox != 0 or cost_giou != 0, "all costs cant be 0" | |
| def forward(self, outputs, targets): | |
| """Performs the matching | |
| Params: | |
| outputs: This is a dict that contains at least these entries: | |
| "pred_logits": Tensor of dim [batch_size, num_queries, num_classes] with the classification logits | |
| "pred_boxes": Tensor of dim [batch_size, num_queries, 4] with the predicted box coordinates | |
| targets: This is a list of targets (len(targets) = batch_size), where each target is a dict containing: | |
| "labels": Tensor of dim [num_target_boxes] (where num_target_boxes is the number of ground-truth | |
| objects in the target) containing the class labels | |
| "boxes": Tensor of dim [num_target_boxes, 4] containing the target box coordinates | |
| Returns: | |
| A list of size batch_size, containing tuples of (index_i, index_j) where: | |
| - index_i is the indices of the selected predictions (in order) | |
| - index_j is the indices of the corresponding selected targets (in order) | |
| For each batch element, it holds: | |
| len(index_i) = len(index_j) = min(num_queries, num_target_boxes) | |
| """ | |
| bs, num_queries = outputs["pred_logits"].shape[:2] | |
| # We flatten to compute the cost matrices in a batch | |
| out_prob = outputs["pred_logits"].flatten(0, 1) # [batch_size * num_queries, num_classes] | |
| # out_prob_bg = 1 - out_prob | |
| # out_prob = torch.cat([out_prob_bg, out_prob], dim = 1) | |
| out_bbox = outputs["pred_boxes"].flatten(0, 1) # [batch_size * num_queries, 4] | |
| # Also concat the target labels and boxes | |
| tgt_bbox = targets["pred_boxes"].flatten(0, 1) # [batch_size * num_target_boxes, 4] | |
| tgt_prob = targets["pred_logits"].flatten(0, 1) # [batch_size * num_target_boxes, num_classes] | |
| # tgt_prob_bg = 1 - tgt_prob | |
| # tgt_prob = torch.cat([tgt_prob_bg, tgt_prob], dim = 1) | |
| # Compute the soft-cross entropy between the predicted token alignment and the GT one for each box | |
| # import pdb | |
| cost_class = out_prob - tgt_prob.transpose(0,1) | |
| cost_class = cost_class.abs() | |
| # Compute the L1 cost between boxes | |
| cost_bbox = torch.cdist(out_bbox, tgt_bbox, p=1) | |
| # Compute the giou cost betwen boxes | |
| # cost_giou = -generalized_box_iou(box_cxcywh_to_xyxy(out_bbox), box_cxcywh_to_xyxy(tgt_bbox)) | |
| cost_giou, _ = box_iou(out_bbox, tgt_bbox) | |
| cost_giou = -cost_giou | |
| # Final cost matrix | |
| C = self.cost_bbox * cost_bbox + self.cost_class * cost_class + self.cost_giou * cost_giou | |
| C = C.view(bs, num_queries, -1).cpu() | |
| C_class = cost_class | |
| C_class = C_class.view(bs, num_queries, -1).cpu() | |
| C_bbox = cost_bbox | |
| C_bbox = C_bbox.view(bs, num_queries, -1).cpu() | |
| #C[torch.isnan(C)] = 0.0 | |
| #C[torch.isinf(C)] = 0.0 | |
| #print(C) | |
| sizes = [tgt_bbox.size(0)] # assum b = 1 | |
| indices = [linear_sum_assignment(c[i]) for i, c in enumerate(C.split(sizes, -1))] | |
| assignment = [(torch.as_tensor(i, dtype=torch.int64), torch.as_tensor(j, dtype=torch.int64)) for i, j in indices] | |
| # calculate the total cost; | |
| assignment = assignment[0] | |
| C = C[0] | |
| C_class = C_class[0] | |
| C_bbox = C_bbox[0] | |
| cost = 0 | |
| selected_entries = [] | |
| cost_class = 0 | |
| cost_bbox = 0 | |
| cost_matched_box = 0 | |
| if self.special: # calculate the difference between boxes | |
| for first_index, second_index in zip(assignment[0], assignment[1]): | |
| if -C[first_index, second_index] > 0.5: | |
| cost += C_class[first_index, second_index] | |
| selected_entries.append(C[first_index, second_index]) | |
| cost_class += C_class[first_index, second_index] | |
| cost_bbox += C_bbox[first_index, second_index] | |
| else: | |
| for first_index, second_index in zip(assignment[0], assignment[1]): | |
| cost += C[first_index, second_index] | |
| selected_entries.append(C[first_index, second_index]) | |
| cost_class += C_class[first_index, second_index] | |
| cost_bbox += C_bbox[first_index, second_index] | |
| print(selected_entries, cost) | |
| return cost, len(selected_entries), selected_entries, cost_class, cost_bbox |