Training pfld_mbv2n_112 gives `ValueError: operands could not be broadcast together`

[sbocconi]

Describe the bug Training with config configs/pfld/pfld_mbv2n_112.py --cfg-options data_root=datasets/meter/ gives ValueError: operands could not be broadcast together with shapes (1638,3) (6,)

Environment Environment you use when bug appears:

1. Python version: 3.10
2. PyTorch Version: torch==2.0.1
3. MMCV Version: 2.0.1
4. EdgeLab Version: na
5. Code you run python tools/train.py configs/pfld/pfld_mbv2n_112.py --cfg-options data_root=datasets/meter/ epochs=10
6. The detailed error

   Traceback (most recent call last):
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/tools/train.py", line 226, in <module>
   main()
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/tools/train.py", line 221, in main
   runner.train()
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmengine/runner/runner.py", line 1777, in train
   model = self.train_loop.run()  # type: ignore
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmengine/runner/loops.py", line 96, in run
   self.run_epoch()
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmengine/runner/loops.py", line 112, in run_epoch
   self.run_iter(idx, data_batch)
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmengine/runner/loops.py", line 128, in run_iter
   outputs = self.runner.model.train_step(
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/detectors/fomo.py", line 99, in train_step
   losses = self._run_forward(data, mode='loss')  # type: ignore
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmengine/model/base_model/base_model.py", line 361, in _run_forward
   results = self(**data, mode=mode)
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
   return forward_call(*args, **kwargs)
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/detectors/fomo.py", line 70, in forward
   return self.loss(inputs, data_samples)
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmdet/models/detectors/single_stage.py", line 78, in loss
   losses = self.bbox_head.loss(x, batch_data_samples)
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/heads/fomo_head.py", line 112, in loss
   loss = self.loss_by_feat(pred, batch_gt_instances, batch_img_metas, batch_gt_instances_ignore)
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/heads/fomo_head.py", line 147, in loss_by_feat
   loss, cls_loss, bg_loss, P, R, F1 = multi_apply(self.lossFunction, preds, target)
   File "/opt/homebrew/Caskroom/miniconda/base/envs/sscma/lib/python3.10/site-packages/mmdet/models/utils/misc.py", line 219, in multi_apply
   return tuple(map(list, zip(*map_results)))
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/heads/fomo_head.py", line 185, in lossFunction
   P, R, F1 = self.get_pricsion_recall_f1(preds, data)
   File "/Users/SB/Projects/Software/Zephyros/Courses/Microcontrollers/ModelAssistant/sscma/models/heads/fomo_head.py", line 231, in get_pricsion_recall_f1
   if site in preds_index:
   ValueError: operands could not be broadcast together with shapes (1638,3) (6,) 

Additional context Running on Mac M2, torch cpu-only, mmcv compiled from source This line in ModelAssistant/sscma/models/heads/fomo_head.py is suspicious:

site = np.concatenate([ti, po], axis=0)

up until that line the sizes where compatible. Changing that line to a np.sum of arrays create another problem later.

[MILK-BIOS]

@sbocconi Thank you for your support! We have fixed the bug and you can try the latest version now!

[sbocconi]

Thanks, it works now!