(self)
| 130 | return self.num_samples * self.num_instances |
| 131 | |
| 132 | def __iter__(self): |
| 133 | indices = torch.randperm(len(self.pids)).tolist() |
| 134 | ret = [] |
| 135 | |
| 136 | for kid in indices: |
| 137 | i = random.choice(self.pid_index[self.pids[kid]]) |
| 138 | # _, i_pid, i_cam = self.data_source[i] |
| 139 | # _, i_pid, i_cam |
| 140 | ret.append(i) |
| 141 | |
| 142 | pid_i = self.index_pid[i] |
| 143 | index = self.pid_index[pid_i] |
| 144 | |
| 145 | select_indexes = No_index(index, i) |
| 146 | if not select_indexes: |
| 147 | continue |
| 148 | if len(select_indexes) >= self.num_instances: |
| 149 | ind_indexes = np.random.choice(select_indexes, size=self.num_instances-1, replace=False) |
| 150 | else: |
| 151 | ind_indexes = np.random.choice(select_indexes, size=self.num_instances-1, replace=True) |
| 152 | |
| 153 | for kk in ind_indexes: |
| 154 | ret.append(index[kk]) |
| 155 | |
| 156 | return iter(ret) |
nothing calls this directly
no test coverage detected