另外,在convert model的时候,要把mmdetection的_base_文件夹导入到mmdeploy的_base_中,把mmdetection的/config/rtmdet导入到mmdeploy的/config中,不然会找不到type.
在尝试使用mmdeploy部署rtmdet ins进行实例分割时，发现网上资源较少，github的问题也有不少未解决，只得自己动手，留作档案。构建时需严格遵循源代码构建流程，从下载到构建，避免第三方库问题。ppl.cv不支持cuda12，构建时需在cuda.cmake中设置与自己显卡对应的flag。使用了最新版的tensorrt和cudnn，版本...
RTMDet-Ins ONNX and TensorRT support. Modification Add rewriter for RTMDetInsHead Fix the rescale of the object detection model class(need to check other models) Inference Results(TRT) Please use configs/mmdet/instance-seg/instance-seg_rtmdet-ins_tensorrt_static-640x640.py mask AP is aligned...
I also encountered this problem. It seems that the loss of RTMDet-ins may exceed the range of fp16 during training and then becomes nan. So I just turned-off the amp mode. Copy link JKellecommentedJun 15, 2023• edited I have two RTMDet-Ins projects that both experience a CUDA OOM...
MMDet 3.0.0rc6 | 大家好,MMDetection 发布了 v3.0.0rc6 版本。 我们积极响应社区需求,在 v3.0.0rc6 中支持了检测器计算 FLOPS 和 TTA(Test Time Augmentation),提供了面向 OpenMMLab 2.0 接口统一的用于模型推理的 DetInferencer ,同时提供了 RTMDet-Ins 的 ONNXRuntime 和 TensorRT 部署教程,此外还支持了许...