Hi, I’m working on an object detection project using one of the models from the NXP eIQ Model Zoo, specifically FastestDet. I’ve trained the model on my own dataset and successfully exported it to ONNX format using the official FastestDet GitHub repository.
Now, I’d like to convert the ONNX model to TFLite format, similar to how the dperson_shufflenetv2.tflite model is provided in the eIQ Model Zoo. However, I couldn’t find any clear guidance on how to structure or convert FastestDet to TFLite properly for use with eIQ.
In the case of NanoDet, there is a recipe.sh script and other well-organized resources in the eIQ Model Zoo to assist with conversion and deployment. Is there anything similar available for FastestDet — like a recipe script, conversion guide, or example configuration? I’d really appreciate any help or recommendations. Thank you!
