Ayush Chaurasia comments

Results 416 comments of


                                            Ayush Chaurasia

the mAP is low with pretrain

This most likely means that your dataset isn't optimal. Try increasing the number of images in the dataset

Confusion matrix and PR curves for testing result!

It's for the val set.

segmentation fault yolov8 during Detection training at the end of epoch before validation

Is this reproducible for you? I see you're running Linux VM on windows. Is the training script running on the linux VM or on windows ?

Integrate trackers

@Laughing-q awesooome!!! @glenn-jocher let's get together this weekend if you're free and finish this up. It should not require any major changes I think

Integrate trackers

Stream=False will most likely cause memory leaks in all cases. This is why we built the streaming mode. When inferencing large video files/streams, the result is accumulated in the memory...

Integrate trackers

@pax7 you can make contributions by forking the repo and submitting a PR.

[Request] Share for mAP improvement for coco-val small size compared to yolov5-7.0

You can see the comparison chart of all models on the readme

object counting in video using yolov8

Thanks for the PR! @glenn-jocher this seems like a good idea. Let's refactor a bit and merge this. What do you think?

object counting in video using yolov8

I'll take over this PR and add some minor updates this weekend.

Yolov8 OBB Task support (Oriented Bounding Box - Object Detection)

@JustasBart This is one of my favorite feature requests! Thanks for the detailed write-up. It perfectly explains the use case. Yes, we might be able to support that but I...