yolov10 predict crashes when augment=True.

THU-MIG / yolov10

YOLOv10: Real-Time End-to-End Object Detection

https://arxiv.org/abs/2405.14458

GNU Affero General Public License v3.0

8k stars 662 forks source link

yolov10 predict crashes when augment=True. #288

Open pierre-bedu-catie opened 1 week ago

pierre-bedu-catie commented 1 week ago

Following recommended installation in a conda env, everything works perfectly fine, except prediction crashes when augment = True specified( TTA). Thanks.

Ashkanbhz commented 1 week ago

I have the same problem.

leonnil commented 1 week ago

Hi, thanks for your interest! I’ve fixed this in the latest codebase. Please update your code.

It’s worth mentioning that since YOLOv10 doesn’t have NMS in the post-processing, it might produce redundant predictions when using Test-Time Augmentation (TTA). So, I suggest being careful when using augment=True to maintain performance.