masouduut94 / volleyball_analytics

This project is designed to display how we can utilize deep learning methods for Sports Data Analytics.
GNU General Public License v2.0
21 stars 6 forks source link

Action detection precision #10

Open haildceu1 opened 3 months ago

haildceu1 commented 3 months ago

Inspiring work! I‘m currently doing similar work except the scenario is on tennis. I'd like to know what action precision you guys can achieve by using YOLO-V8, considering that no temporal features are involved. Also, have you guys heard of SwingVision? What algorithm do you think this app uses cause it looks pretty amazing. Much appreciation!

masouduut94 commented 3 months ago

Have a look at these results:

https://github.com/masouduut94/volleyball_analytics/wiki/Deep-Learning-Models#action-detection

haildceu1 commented 3 months ago

Have a look at these results:

https://github.com/masouduut94/volleyball_analytics/wiki/Deep-Learning-Models#action-detection

Thank you. I'm also wondering what kind of GPU do you run the algorithm on. As far as I know, VideoMAE is quite large.

masouduut94 commented 3 months ago

I trained the VideoMAE model with RTX 3090 24 GB. It doesn't need big GPUs

On Sat, Jul 27, 2024, 10:42 AM Zhuang Yiwen @.***> wrote:

Have a look at these results:

https://github.com/masouduut94/volleyball_analytics/wiki/Deep-Learning-Models#action-detection

Thank you. I'm also wondering what kind of GPU do you run the algorithm on. As far as I know, VideoMAE is quite large.

— Reply to this email directly, view it on GitHub https://github.com/masouduut94/volleyball_analytics/issues/10#issuecomment-2253868915, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFB4FCDS3GHINQASZIP7V2DZONB5PAVCNFSM6AAAAABLOQCELKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENJTHA3DQOJRGU . You are receiving this because you commented.Message ID: @.***>