CenterNet icon indicating copy to clipboard operation
CenterNet copied to clipboard

Object detection, 3D detection, and pose estimation using center point detection:

Results 152 CenterNet issues
Sort by recently updated
recently updated
newest added

TLDR: This fix leads to better performance in rotation prediction. First of all, thank you very much for publicating your work. It was super helpful for my work as well....

![XVYB1@~CB 41RS@364ZAO66](https://user-images.githubusercontent.com/105423291/192739255-82f9bd9f-9184-4023-be9f-3ea584f3d91f.png) 教教孩子吧 在网上看了很多教程 说在文件目录下放一个__init__.py 再把文件夹设置为source root 都尝试了 但是还是定位不到logger 有知道的吗 帮忙解答一下!

I did python3 src/demo.py ctdet --demo images --load_model models/ctdet_coco_dla_2x.pth and python3 src/demo.py multi_pose --demo images --load_model models/multi_pose_dla_3x.pth Using centernet docker image I got error THCudaCheck FAIL file=/pytorch/aten/src/THCUNN/generic/Threshold.cu line=34 error=48 :...

(measure) C:\Users\IFL\Desktop\Liu\Detector\CenterNet\src>python main.py ctdet --exp_id 640 --batch_size 4 --dataset corner --input_res 640 --num_epochs 150 --val_intervals 10 Fix size testing. training chunk_sizes: [4] The output will be saved to C:\Users\IFL\Desktop\Liu\Detector\CenterNet\src\lib\..\..\exp\ctdet\640 heads...

what's the meaning of "--master_batch"?

我参考dlav的结构把vovnet和resnest加入进了训练,发现新的两个backbone在各种深度下Loss都不能很好地下降,训练coco数据集总的loss在8左右

When training the **pascal voc** dataset, the offset loss down from 0.28 firstly, regardless of many training epoches, and finally around 0.2 all the time。 So anyone ran into this?...