-
数据集是HRSC2016 采用配置文件cfgs_res50_dota_v1时loss可以正常收敛,当采用cfgs_res50_dota_v5时cls loss始终在1-1.2之间,reg loss 出现nan,tensorboard中gtboxes_h_gpu、gtboxes_r_gpu_正常有框,positive_anchor也是正常有anchor,但是final_detection_gpu_0…
-
Currently the round slider supports the single dial option only (that means starting 0 degree to 360 degree). Similar. need to provide the multiple dial option. So that any one can drag the handle aro…
-
Hello, thanks much for your very nice repository.
I got some models successfully working in the closed loop simulation.
I also have the output images and can generate the videos via generate_video.p…
-
### Description
```modelica
model test_Poisson
block PoissonProcess "Generate fixed magnitude events with exponentially distributed interval"
import generator = Modelica.Math.Random.Generators…
-
Has anyone reproduced the VAD experiment results?
(Bench2DriveZoo/adzoo/vad/configs/VAD/VAD_base_e2e_b2d.py+Bench2DriveZoo/ckpts/vad_b2d_base.pth)
If so, could you please share the commit ID and t…
-
my machine :RTX3090 ,Computer memory 32G
![image](https://github.com/user-attachments/assets/a5a899fc-7007-4827-88d4-fb72c1d8aa83)
-
Your work has set a benchmark in closed-loop evaluation for end-to-end autonomous driving methods, offering researchers a unified and fair evaluation framework that facilitates easier model assessment…
-
Once we have this working, and easily configurable, I would put something like this in the README file?
1. Fork the repository on GitHub
2. Turn on TravisCI testing for the forked repository
3. C…
-
When I replay the log video by doing client.replay_file('log_file.log', 0,0,0), it shows a BEV of an intersection, with cars moving but quickly disappearing. It does not follow an ego vehicle's trajec…
-
Dear Bench2Drive Developers,
The annotations in the Bench2Drive dataset provide the following items:
```
- x_command_far # farther waypoint x in world coordinates
- y_command_far # farther waypo…