Laughing

@SyGoing please check your dataset and try to delete the old .cache files to force re-cache.

@SyGoing I tested again and the existed cache files works fine to me. could you try coco128.yaml to see if this issue still happen?

@FreedomLiX the image would be converted to 640x640 without any padding if stretch=True.

@yunxi1 cocoapi evaluation will automatically be started once this `is_coco` is True. [https://github.com/ultralytics/ultralytics/blob/d0b616e41e827e1ac65ee99da1339e707a984989/ultralytics/yolo/v8/detect/val.py#L47](https://github.com/ultralytics/ultralytics/blob/d0b616e41e827e1ac65ee99da1339e707a984989/ultralytics/yolo/v8/detect/val.py#L47)

@AyushExel sure!

@AyushExel all the `Results(orig_shape=self.orig_shape)` should update to `Results(orig_img=self.orig_img)` I think. [https://github.com/ultralytics/ultralytics/blob/41c7e894431fe8962cf6f278c36757f9e6626c37/ultralytics/yolo/engine/results.py#L62](https://github.com/ultralytics/ultralytics/blob/41c7e894431fe8962cf6f278c36757f9e6626c37/ultralytics/yolo/engine/results.py#L62) And how about moving `names` to `def visualize(...name=None)` instead of making it as a member of `Results`? since `names`...

@AyushExel also we probably need more Results-related operation tests in our test script, like `results.cpu()`, `results.numpy()` and so on.

@kyoko9000 hey can you try it again with the latest version 8.0.5? thanks!

@kyoko9000 oh just `pip install --upgrade ultralytics` should work.

@AI-P-K the output of predict is a python generator for now. you can try this to get your results: ```python for output in results: pass ``` the output is a...