slowfast detection problem

[yhArcadia]

The doc issue

Using version 1.0 of mmaction2, I changed the object detection phase to yolov3, run demo_spatiotemporal_det.py ,using the trained yolov3 weights and slowfast weights，encountered this error，the progress bar of behavior recognition did not grow， and the output video does not have a detection box or label. How should I solve this problem?

使用mmaction2的1.0版本， 目标检测阶段换成yolov3之后，用自己训练的yolov3权重和slowfast权重，运行demo_spatiotemporal_det.py遇到了如下的问题，行为识别的进度条没有增长，且输出的视频没有检测框和标签，跟原视频一样，请问该如何处理呢？

Suggest a potential alternative/fix

No response

[hukkai]

@yhArcadia Did your yolov3 detector returned the same format bbox as the original detector?

[dreamrain123]

请问，问题解决了吗 我遇到和你同样的问题了