NN Model converter
Budget: $45 – $55 USD
Create a python script that does the following:
• Convert pytorch yolov7 and MobileNet V2/3 models to ONNX
• Convert same models from ONNX to tflight (Edge TPU)
◦ argument to enable quantization (Post-training quantization)
• Convert same models from ONNX to RKNN (RK NPU)
◦ argument to enable quantization (Post-training quantization)
Follow all best practices and performance considerations. The output networks should keep the accurance and run the NPU/TPU as fast as they should. You will be given access to a device with RK NPU and Coral EdgeTPU to test the script and the output models.
The price is fixed. A person with the right experience should do in timely manner.
Use keywords “ONNX Guru”or your application wont be considered!
• Convert pytorch yolov7 and MobileNet V2/3 models to ONNX
• Convert same models from ONNX to tflight (Edge TPU)
◦ argument to enable quantization (Post-training quantization)
• Convert same models from ONNX to RKNN (RK NPU)
◦ argument to enable quantization (Post-training quantization)
Follow all best practices and performance considerations. The output networks should keep the accurance and run the NPU/TPU as fast as they should. You will be given access to a device with RK NPU and Coral EdgeTPU to test the script and the output models.
The price is fixed. A person with the right experience should do in timely manner.
Use keywords “ONNX Guru”or your application wont be considered!