diff options
author | Jordan Gong <jordan.gong@protonmail.com> | 2021-02-20 14:19:30 +0800 |
---|---|---|
committer | Jordan Gong <jordan.gong@protonmail.com> | 2021-02-20 14:31:13 +0800 |
commit | 820d3dec284f38e6a3089dad5277bc3f6c5123bf (patch) | |
tree | d9ba88fdc429ab14ddb5cd4278215dbe5df4a424 /utils/triplet_loss.py | |
parent | 4aa9044122878a8e2b887a8b170c036983431559 (diff) |
Separate triplet loss from model
Diffstat (limited to 'utils/triplet_loss.py')
-rw-r--r-- | utils/triplet_loss.py | 58 |
1 files changed, 50 insertions, 8 deletions
diff --git a/utils/triplet_loss.py b/utils/triplet_loss.py index 954def2..0df2188 100644 --- a/utils/triplet_loss.py +++ b/utils/triplet_loss.py @@ -11,6 +11,25 @@ class BatchAllTripletLoss(nn.Module): def forward(self, x, y): p, n, c = x.size() + dist = self._batch_distance(x) + positive_negative_dist = self._hard_distance(dist, y, p, n) + all_loss = F.relu(self.margin + positive_negative_dist).view(p, -1) + parted_loss_mean = self._none_zero_parted_mean(all_loss) + + return parted_loss_mean + + @staticmethod + def _hard_distance(dist, y, p, n): + hard_positive_mask = y.unsqueeze(1) == y.unsqueeze(2) + hard_negative_mask = y.unsqueeze(1) != y.unsqueeze(2) + all_hard_positive = dist[hard_positive_mask].view(p, n, -1, 1) + all_hard_negative = dist[hard_negative_mask].view(p, n, 1, -1) + positive_negative_dist = all_hard_positive - all_hard_negative + + return positive_negative_dist + + @staticmethod + def _batch_distance(x): # Euclidean distance p x n x n x_squared_sum = torch.sum(x ** 2, dim=2) x1_squared_sum = x_squared_sum.unsqueeze(2) @@ -20,17 +39,40 @@ class BatchAllTripletLoss(nn.Module): F.relu(x1_squared_sum - 2 * x1_times_x2_sum + x2_squared_sum) ) - hard_positive_mask = y.unsqueeze(1) == y.unsqueeze(2) - hard_negative_mask = y.unsqueeze(1) != y.unsqueeze(2) - all_hard_positive = dist[hard_positive_mask].view(p, n, -1, 1) - all_hard_negative = dist[hard_negative_mask].view(p, n, 1, -1) - positive_negative_dist = all_hard_positive - all_hard_negative - all_loss = F.relu(self.margin + positive_negative_dist).view(p, -1) + return dist + @staticmethod + def _none_zero_parted_mean(all_loss): # Non-zero parted mean non_zero_counts = (all_loss != 0).sum(1) parted_loss_mean = all_loss.sum(1) / non_zero_counts parted_loss_mean[non_zero_counts == 0] = 0 - loss = parted_loss_mean.mean() - return loss + return parted_loss_mean + + +class JointBatchAllTripletLoss(BatchAllTripletLoss): + def __init__( + self, + hpm_num_parts: int, + margins: tuple[float, float] = (0.2, 0.2) + ): + super().__init__() + self.hpm_num_parts = hpm_num_parts + self.margin_hpm, self.margin_pn = margins + + def forward(self, x, y): + p, n, c = x.size() + + dist = self._batch_distance(x) + positive_negative_dist = self._hard_distance(dist, y, p, n) + hpm_part_loss = F.relu( + self.margin_hpm + positive_negative_dist[:self.hpm_num_parts] + ).view(self.hpm_num_parts, -1) + pn_part_loss = F.relu( + self.margin_pn + positive_negative_dist[self.hpm_num_parts:] + ).view(p - self.hpm_num_parts, -1) + all_loss = torch.cat((hpm_part_loss, pn_part_loss)).view(p, -1) + parted_loss_mean = self._none_zero_parted_mean(all_loss) + + return parted_loss_mean |