Laughing

@Auggie3 oh I see it's about the corrupt labels. I've reproduced the issue and made a PR #902 to fix it. :) Thanks!

@MuhammadSibtain5099 hi I think it's already supported. :) ```bash yolo export model=yolov8n.pt format=tflite ```

@hariouat refer to here [https://github.com/ultralytics/ultralytics/blob/09265b17d7bbe0f7d45d25086347fe8197cb6970/ultralytics/models/v8/seg/yolov8l-seg.yaml#L25](https://github.com/ultralytics/ultralytics/blob/09265b17d7bbe0f7d45d25086347fe8197cb6970/ultralytics/models/v8/seg/yolov8l-seg.yaml#L25) just give the correct index of the feature map.

@hariouat can you show me how you implement the `Multiply` module?

@hariouat 1. you should add these two lines in ultralytics/nn/tasks.py:  2. modify your implementation to: ```python class Multiply(nn.Module): def __init__(self): super(Multiply, self).__init__() def forward(self, x): return torch.mul(x[0], x[1]) ```...

@mehran66 hi the issue had been solved in this PR #651 . please make sure you're using the latest version. :) @saka3324 I noticed you opened a same issue #618...

@nikbobo yeah we also don't want to introduce more dependencies for now so reid models won't be implemented in this PR. But it's on todo list, we'd like to implement...

@glenn-jocher yeah it's ready to merge. :) actually the `im0.copy()` is a small patch to fix overlapped drawing box when the source is stream:  like the `im0s.copy()` in v5...

@glenn-jocher this way to fix `im0` should be better. [https://github.com/ultralytics/ultralytics/pull/788/commits/14871425118ec863f7b3e1c1337d68dac7e5ed23](https://github.com/ultralytics/ultralytics/pull/788/commits/14871425118ec863f7b3e1c1337d68dac7e5ed23)

@mikel-brostrom yeah, but not in this PR. :))