Open CharlesGaydon opened 2 months ago
Note: maybe https://github.com/IGNF/myria3d/pull/124 has corrected this since the computation of the IoU is now performed using a different function. Feel free to close if you do not observe this behavior again.
Observed in DDP mode with 2x3 GPUs (not tested on single node training).
This is surprising, because the per-class IoUs are computed with a bespoke method from the confusion matrix of the MultiClassJaccardIndex object that computes mean IoU
https://github.com/IGNF/myria3d/blob/d52810eb6024c6b4dbb7816ff3be83fb1dd732b0/myria3d/models/model.py#L190
Mean IoU:![image](https://github.com/IGNF/myria3d/assets/11660435/91288e20-7ef4-4fbf-b177-de9e0e4d7937)
Per-class:![image](https://github.com/IGNF/myria3d/assets/11660435/853a2983-0310-433b-874e-690c2edab162)
Might be linked to https://github.com/IGNF/myria3d/issues/108