from loguru import logger import torch import torch.nn as nn class ASpanLoss(nn.Module): def __init__(self, config): super().__init__() self.config = config # config under the global namespace self.loss_config = config["aspan"]["loss"] self.match_type = self.config["aspan"]["match_coarse"]["match_type"] self.sparse_spvs = self.config["aspan"]["match_coarse"]["sparse_spvs"] self.flow_weight = self.config["aspan"]["loss"]["flow_weight"] # coarse-level self.correct_thr = self.loss_config["fine_correct_thr"] self.c_pos_w = self.loss_config["pos_weight"] self.c_neg_w = self.loss_config["neg_weight"] # fine-level self.fine_type = self.loss_config["fine_type"] def compute_flow_loss(self, coarse_corr_gt, flow_list, h0, w0, h1, w1): # coarse_corr_gt:[[batch_indices],[left_indices],[right_indices]] # flow_list: [L,B,H,W,4] loss1 = self.flow_loss_worker( flow_list[0], coarse_corr_gt[0], coarse_corr_gt[1], coarse_corr_gt[2], w1 ) loss2 = self.flow_loss_worker( flow_list[1], coarse_corr_gt[0], coarse_corr_gt[2], coarse_corr_gt[1], w0 ) total_loss = (loss1 + loss2) / 2 return total_loss def flow_loss_worker(self, flow, batch_indicies, self_indicies, cross_indicies, w): bs, layer_num = flow.shape[1], flow.shape[0] flow = flow.view(layer_num, bs, -1, 4) gt_flow = torch.stack([cross_indicies % w, cross_indicies // w], dim=1) total_loss_list = [] for layer_index in range(layer_num): cur_flow_list = flow[layer_index] spv_flow = cur_flow_list[batch_indicies, self_indicies][:, :2] spv_conf = cur_flow_list[batch_indicies, self_indicies][ :, 2: ] # [#coarse,2] l2_flow_dis = (gt_flow - spv_flow) ** 2 # [#coarse,2] total_loss = spv_conf + torch.exp(-spv_conf) * l2_flow_dis # [#coarse,2] total_loss_list.append(total_loss.mean()) total_loss = torch.stack(total_loss_list, dim=-1) * self.flow_weight return total_loss def compute_coarse_loss(self, conf, conf_gt, weight=None): """Point-wise CE / Focal Loss with 0 / 1 confidence as gt. Args: conf (torch.Tensor): (N, HW0, HW1) / (N, HW0+1, HW1+1) conf_gt (torch.Tensor): (N, HW0, HW1) weight (torch.Tensor): (N, HW0, HW1) """ pos_mask, neg_mask = conf_gt == 1, conf_gt == 0 c_pos_w, c_neg_w = self.c_pos_w, self.c_neg_w # corner case: no gt coarse-level match at all if not pos_mask.any(): # assign a wrong gt pos_mask[0, 0, 0] = True if weight is not None: weight[0, 0, 0] = 0.0 c_pos_w = 0.0 if not neg_mask.any(): neg_mask[0, 0, 0] = True if weight is not None: weight[0, 0, 0] = 0.0 c_neg_w = 0.0 if self.loss_config["coarse_type"] == "cross_entropy": assert ( not self.sparse_spvs ), "Sparse Supervision for cross-entropy not implemented!" conf = torch.clamp(conf, 1e-6, 1 - 1e-6) loss_pos = -torch.log(conf[pos_mask]) loss_neg = -torch.log(1 - conf[neg_mask]) if weight is not None: loss_pos = loss_pos * weight[pos_mask] loss_neg = loss_neg * weight[neg_mask] return c_pos_w * loss_pos.mean() + c_neg_w * loss_neg.mean() elif self.loss_config["coarse_type"] == "focal": conf = torch.clamp(conf, 1e-6, 1 - 1e-6) alpha = self.loss_config["focal_alpha"] gamma = self.loss_config["focal_gamma"] if self.sparse_spvs: pos_conf = ( conf[:, :-1, :-1][pos_mask] if self.match_type == "sinkhorn" else conf[pos_mask] ) loss_pos = -alpha * torch.pow(1 - pos_conf, gamma) * pos_conf.log() # calculate losses for negative samples if self.match_type == "sinkhorn": neg0, neg1 = conf_gt.sum(-1) == 0, conf_gt.sum(1) == 0 neg_conf = torch.cat( [conf[:, :-1, -1][neg0], conf[:, -1, :-1][neg1]], 0 ) loss_neg = -alpha * torch.pow(1 - neg_conf, gamma) * neg_conf.log() else: # These is no dustbin for dual_softmax, so we left unmatchable patches without supervision. # we could also add 'pseudo negtive-samples' pass # handle loss weights if weight is not None: # Different from dense-spvs, the loss w.r.t. padded regions aren't directly zeroed out, # but only through manually setting corresponding regions in sim_matrix to '-inf'. loss_pos = loss_pos * weight[pos_mask] if self.match_type == "sinkhorn": neg_w0 = (weight.sum(-1) != 0)[neg0] neg_w1 = (weight.sum(1) != 0)[neg1] neg_mask = torch.cat([neg_w0, neg_w1], 0) loss_neg = loss_neg[neg_mask] loss = ( c_pos_w * loss_pos.mean() + c_neg_w * loss_neg.mean() if self.match_type == "sinkhorn" else c_pos_w * loss_pos.mean() ) return loss # positive and negative elements occupy similar propotions. => more balanced loss weights needed else: # dense supervision (in the case of match_type=='sinkhorn', the dustbin is not supervised.) loss_pos = ( -alpha * torch.pow(1 - conf[pos_mask], gamma) * (conf[pos_mask]).log() ) loss_neg = ( -alpha * torch.pow(conf[neg_mask], gamma) * (1 - conf[neg_mask]).log() ) if weight is not None: loss_pos = loss_pos * weight[pos_mask] loss_neg = loss_neg * weight[neg_mask] return c_pos_w * loss_pos.mean() + c_neg_w * loss_neg.mean() # each negative element occupy a smaller propotion than positive elements. => higher negative loss weight needed else: raise ValueError( "Unknown coarse loss: {type}".format( type=self.loss_config["coarse_type"] ) ) def compute_fine_loss(self, expec_f, expec_f_gt): if self.fine_type == "l2_with_std": return self._compute_fine_loss_l2_std(expec_f, expec_f_gt) elif self.fine_type == "l2": return self._compute_fine_loss_l2(expec_f, expec_f_gt) else: raise NotImplementedError() def _compute_fine_loss_l2(self, expec_f, expec_f_gt): """ Args: expec_f (torch.Tensor): [M, 2] expec_f_gt (torch.Tensor): [M, 2] """ correct_mask = ( torch.linalg.norm(expec_f_gt, ord=float("inf"), dim=1) < self.correct_thr ) if correct_mask.sum() == 0: if ( self.training ): # this seldomly happen when training, since we pad prediction with gt logger.warning("assign a false supervision to avoid ddp deadlock") correct_mask[0] = True else: return None flow_l2 = ((expec_f_gt[correct_mask] - expec_f[correct_mask]) ** 2).sum(-1) return flow_l2.mean() def _compute_fine_loss_l2_std(self, expec_f, expec_f_gt): """ Args: expec_f (torch.Tensor): [M, 3] expec_f_gt (torch.Tensor): [M, 2] """ # correct_mask tells you which pair to compute fine-loss correct_mask = ( torch.linalg.norm(expec_f_gt, ord=float("inf"), dim=1) < self.correct_thr ) # use std as weight that measures uncertainty std = expec_f[:, 2] inverse_std = 1.0 / torch.clamp(std, min=1e-10) weight = ( inverse_std / torch.mean(inverse_std) ).detach() # avoid minizing loss through increase std # corner case: no correct coarse match found if not correct_mask.any(): if ( self.training ): # this seldomly happen during training, since we pad prediction with gt # sometimes there is not coarse-level gt at all. logger.warning("assign a false supervision to avoid ddp deadlock") correct_mask[0] = True weight[0] = 0.0 else: return None # l2 loss with std flow_l2 = ((expec_f_gt[correct_mask] - expec_f[correct_mask, :2]) ** 2).sum(-1) loss = (flow_l2 * weight[correct_mask]).mean() return loss @torch.no_grad() def compute_c_weight(self, data): """compute element-wise weights for computing coarse-level loss.""" if "mask0" in data: c_weight = ( data["mask0"].flatten(-2)[..., None] * data["mask1"].flatten(-2)[:, None] ).float() else: c_weight = None return c_weight def forward(self, data): """ Update: data (dict): update{ 'loss': [1] the reduced loss across a batch, 'loss_scalars' (dict): loss scalars for tensorboard_record } """ loss_scalars = {} # 0. compute element-wise loss weight c_weight = self.compute_c_weight(data) # 1. coarse-level loss loss_c = self.compute_coarse_loss( data["conf_matrix_with_bin"] if self.sparse_spvs and self.match_type == "sinkhorn" else data["conf_matrix"], data["conf_matrix_gt"], weight=c_weight, ) loss = loss_c * self.loss_config["coarse_weight"] loss_scalars.update({"loss_c": loss_c.clone().detach().cpu()}) # 2. fine-level loss loss_f = self.compute_fine_loss(data["expec_f"], data["expec_f_gt"]) if loss_f is not None: loss += loss_f * self.loss_config["fine_weight"] loss_scalars.update({"loss_f": loss_f.clone().detach().cpu()}) else: assert self.training is False loss_scalars.update({"loss_f": torch.tensor(1.0)}) # 1 is the upper bound # 3. flow loss coarse_corr = [data["spv_b_ids"], data["spv_i_ids"], data["spv_j_ids"]] loss_flow = self.compute_flow_loss( coarse_corr, data["predict_flow"], data["hw0_c"][0], data["hw0_c"][1], data["hw1_c"][0], data["hw1_c"][1], ) loss_flow = loss_flow * self.flow_weight for index, loss_off in enumerate(loss_flow): loss_scalars.update( {"loss_flow_" + str(index): loss_off.clone().detach().cpu()} ) # 1 is the upper bound conf = data["predict_flow"][0][:, :, :, :, 2:] layer_num = conf.shape[0] for layer_index in range(layer_num): loss_scalars.update( { "conf_" + str(layer_index): conf[layer_index] .mean() .clone() .detach() .cpu() } ) # 1 is the upper bound loss += loss_flow.sum() # print((loss_c * self.loss_config['coarse_weight']).data,loss_flow.data) loss_scalars.update({"loss": loss.clone().detach().cpu()}) data.update({"loss": loss, "loss_scalars": loss_scalars})