yzfly / SimpleTreeHeight

A simple algorithm for extracting tree height in sparse scene from point cloud data.
MIT License
10 stars 1 forks source link

JHMDB-21 training gt #3

Closed Holmes-GU closed 4 years ago

Holmes-GU commented 4 years ago

Hi, I am a student from SJTU. Would you like to share JHMDB-21 training gt? Thank you very much

Zhihao Gu

yzfly commented 4 years ago

Hi, I'm also a student, I’m from UESTC. Actually,I have shared it. Please reference to issues25

image

Please note that this is generated by myself, accuracy cannot be guaranteed. If you can't download it, tell me your email and I'll try to give you a copy.

I use it and get similar results as YOWO. If you can't reproduce the results, please select your hyperparameter carefully. With some improvements, I got better results than YOWO on both jhmdb21 and ucf24.

Holmes-GU commented 4 years ago

Ok, my mail is gzhmsy@126.com,one more question, when you evaluate video-map on ucf and jhmdb, where do you find the testlist-video files? I failed to find them. Yours

SJTU 顾智浩 邮箱:gzhmsy@126.com

签名由 网易邮箱大师 定制

On 08/11/2020 05:57, yzfly wrote:

Hi, I'm also a student, I’m from UESTC. Actually,I have shared it. Please reference to issues25

Please note that this is generated by myself, accuracy cannot be guaranteed. If you can't download it, tell me your email and I'll try to give you a copy.

I use it and get similar results as YOWO. If you can't reproduce the results, please select your hyperparameter carefully. With some improvements, I got better results than YOWO on both jhmdb21 and ucf24.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

r09921135 commented 2 years ago

@yzfly Hi! Would you mind to share the hyperparameters you choose or what kinds of improvements you made? I've been struggling to reproduce the results... very looking forward to your respond! Thank you very much!

yzfly commented 2 years ago

It's been too long to remember. Maybe you can refer to our paper https://openaccess.thecvf.com/content/ACCV2020W/MMHAU/html/Liu_Real-time_spatio-temporal_action_localization_via_learning_motion_representation_ACCVW_2020_paper.html