Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Training only Human class, map with Pedestrian class in Kitti but result is 0.0000 #1698

Open
thanhpt93 opened this issue Jan 5, 2025 · 1 comment

Comments

@thanhpt93
Copy link

thanhpt93 commented Jan 5, 2025

Hi everyone,
I have issue that when I train only class is human, and I mapping with pedestrian class in Kitti. But result is not as expect, all is 0.0000 as the bellow:

2025-01-05 10:33:55,387 INFO *************** Performance of EPOCH 10 *****************
2025-01-05 10:33:55,387 INFO Generate label finished(sec_per_example: 0.6781 second).
2025-01-05 10:33:55,387 INFO recall_roi_0.5: 0.775510
2025-01-05 10:33:55,387 INFO recall_rcnn_0.5: 0.843537
2025-01-05 10:33:55,387 INFO recall_roi_0.7: 0.442177
2025-01-05 10:33:55,387 INFO recall_rcnn_0.7: 0.455782
2025-01-05 10:33:55,387 INFO Average predicted number of objects(78 samples): 1.962
/home/thanhpt/miniconda3/envs/3d/lib/python3.8/site-packages/numba/cuda/dispatcher.py:536: NumbaPerformanceWarning: Grid size 9 will likely result in GPU under-utilization due to low occupancy.
warn(NumbaPerformanceWarning(msg))
2025-01-05 10:33:55,414 INFO Pedestrian [email protected], 0.50, 0.50:
bbox AP:0.0000, 0.0000, 0.0000
bev AP:0.0000, 0.0000, 0.0000
3d AP:0.0000, 0.0000, 0.0000
aos AP:0.00, 0.00, 0.00
Pedestrian [email protected], 0.50, 0.50:
bbox AP:0.0000, 0.0000, 0.0000
bev AP:0.0000, 0.0000, 0.0000
3d AP:0.0000, 0.0000, 0.0000
aos AP:0.00, 0.00, 0.00
Pedestrian [email protected], 0.25, 0.25:
bbox AP:0.0000, 0.0000, 0.0000
bev AP:0.0000, 0.0000, 0.0000
3d AP:0.0000, 0.0000, 0.0000
aos AP:0.00, 0.00, 0.00
Pedestrian [email protected], 0.25, 0.25:
bbox AP:0.0000, 0.0000, 0.0000
bev AP:0.0000, 0.0000, 0.0000
3d AP:0.0000, 0.0000, 0.0000
aos AP:0.00, 0.00, 0.00

For more detail, I have pretrain is PV-RCNN. VOXEL SIZE is (0.1, 0.1, 0.1). Custom data has been checked annotation is correct
I wonder if anyone has encountered this. Hope to hear from you guys!

@Petros626
Copy link

Petros626 commented Jan 8, 2025

maybe from Pedestrians there aren't so many samples contained? What is the number Pedestrians for gt_sampling in the .yaml? How does your training loss look like?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants