NOTE: code based on WXinlong/SOLO

add support to TensorRT inference

• onnxruntime
• tensorRT
• full_dims and dynamic shape
• postprocess without torch

git clone https://github.com/zhangjinsong3/SOLOv2.tensorRT

• pytorch 1.4+
• torchvision 0.5+
• gcc 4.9+
• tensorRT 7.2.1
• onnx-tensorrt 7.2
• onnx 1.8.0
• onnxruntime 1.6.0

• set ONNX_EXPORT=False in mmdet.deploy_params.py
• Training according to the steps in SOLO🔗

• set ONNX_EXPORT=True in mmdet.deploy_params.py

• cd deploy

• run inference on onnxruntime & tensorRT

# run on pytorch
python inference_demo.py

# Convert pth to onnx model, then run on onnxrt & trt
# SOLOv2 light R34
python onnx_exporter_solov2.py ../configs/solov2/solov2_light_448_r34_fpn_8gpu_3x.py  weights/SOLOv2_light_R34.onnx --checkpoint ../weights/SOLOv2_LIGHT_448_R34_3x.pth --shape 448 672
python inference_on_onnxrt_trt_solov2.py