David-19940718

@clemente620 start your id from [0, 1, ...] which corresponding to your modify classes, such as classes = ['__background__', 'class1', 'class2'], then you should set ids = [0, 1, 2]....

@mrkind Hello, mrkind, you can add this code in /path/to/CentetNet/src/lib/detectors/ctdet.py debugger.save_all_imgs(path='/home/username/path/to/CenterNet/output/', genID=True) 1- Put this code in the last line; 2- Create a new file name "output" and put it...

I think you can check your output, it may be clear.

@mrkind I found a solution to save the results(bounding box, predicted_id with the corresponding score ): [https://blog.csdn.net/weixin_43509263/article/details/100799415](url)

The same as issue for me!

It worked! Thank you so much.---- Replied Message ***@***.***>Date09/21/2022 23:54 ***@***.***> CcJack ***@***.******@***.***>SubjectRe: [wufan-tb/yolo_slowfast] Please help me solve the error orz (Issue #10) I have tried it and succeed.But I...

请问下我导出onnx推理后，得到一个300*6的矩阵，这6个值分别是[score, class, cx, cy, w, h]吗？你们那边有对应的后处理解析代码能提供下吗？我解析出来的框对不上。类别和置信度倒是对了。 

> * 同样的脚本和图片，并不能复现你说的问题（下面是推理保存的图片），，确保你代码和模型是最新的 > > `python tools/infer.py -c configs/rtdetr/rtdetr_r50vd_6x_coco.yml -o weights=https://bj.bcebos.com/v1/paddledet/models/rtdetr_r50vd_6x_coco.pdparams --infer_img=bus.jpg` > >  您好，我刚重新下载了最新的代码，还是一样得不到结果。请问这跟CUDNN版本不匹配有关系吗？

您好，我将绘制阈值调低到0.01，出来的结果是这样的： 

> @David-19940718 ONNX和TensorRT目前正在测试过程中，马上会更新。 hello，请问这个计划什么时间会发布？