Hey there community,
I have been trying hard to run a CNN + LSTM architecture on the Hailo 8L with raspberry pi5, I chose Yolov9c model, then converted it to ONNX with only the feature embedding as the output, name ‘lstm_feature’. I then converted it to .hef format for execution on, the hailo8L module.
I was take some help from our friend Claude 4 Sonnet, as was able to get a script to run the thing, however the script showed a lot of errors and troubleshooting them pointed me to the fact that this custom experiment with the yolov9c is not working.
Basically the hailonet was not executing the hef, properly and simply passing the collected frame as the output. Due to which there was no proper input for the LSTM.
Can someone point me to some genuine resource or, help me out, cause I literally feel hopeless.