RangiLyu / nanodet

NanoDet-Plus⚡Super fast and lightweight anchor-free object detection model. 🔥Only 980 KB(int8) / 1.8MB (fp16) and run 97FPS on cellphone🔥
Apache License 2.0
5.71k stars 1.04k forks source link

Why did this error occur during verification? #540

Open LiuShiChen-Chn opened 11 months ago

LiuShiChen-Chn commented 11 months ago

output = self._evaluation_step(kwargs) File "/site-packages/pytorch_lightning/loops/epoch/evaluation_epoch_loop.py", line 234, in _evaluation_step output = self.trainer._call_strategy_hook(hook_name, kwargs.values()) File "/site-packages/pytorch_lightning/trainer/trainer.py", line 1494, in _call_strategy_hook output = fn(args, kwargs) File "/site-packages/pytorch_lightning/strategies/strategy.py", line 390, in validation_step return self.model.validation_step(*args, **kwargs) File "/nanodet/nanodet/trainer/task.py", line 140, in validation_step dets = self.model.head.post_process(preds, batch) File "/nanodet/model/head/nanodet_plus_head.py", line 440, in post_process det_bboxes[:, :4] = warp_boxes( File "/nanodet/nanodet/data/transform/warp.py", line 212, in warp_boxes xy = (xy[:, :2] / xy[:, 2:3]).reshape(n, 8) # rescale ValueError: cannot reshape array of size 18 into shape (3,8)