Is yolov4-csp batched inference supported?

[LukeAI]

thanks for this great project! I wanted to ask if it's possible to get batch size 4 inference working with yolov4-csp ?

I got yolov4-csp tensorrt inference working fine with the default batch_size=1 but if I set batch_size=3 or 4 inside export_onnx.py then on export I get:

python3 export_onnx.py --cfg_file cfg/y
olov4-csp.cfg --weights_file yolov4-csp.weights --output_file yolov4-csp.onnx                        
Layer of type yolo not supported, skipping ONNX node generation.                                     
Layer of type yolo not supported, skipping ONNX node generation.                                     
Layer of type yolo not supported, skipping ONNX node generation. 

and when running inference (with batch_size=3 set in config-yolov4-csp-3.yaml )

[01/15/2021-09:35:02] [E] [TRT] Parameter check failed at: engine.cpp::execute::806, condition: batchSize > 0 && batchSize <= mEngine.getMaxBatchSize()
Inference take: 0.05127 ms.
execute success
device2host
post process
Segmentation fault (core dumped)

[justinkay]

Hi @LukeAI , how did you resolve this issue?

[LukeAI]

I didn't - I just use yolov4-large-p5 at a slightly lower resolution