I have used the Tensorflow Object Detection API to train models on a custom dataset. The tensorflow object detection API also allows evaluating the trained models on a test set and gives results in the COCO eval format. An example output from the evaluation can be seen here:
Evaluation output from Tensorflow Object Detection API reporting the MSCOCO metrics
In this image, there are results such as mAP @ 0.5 IoU and mAP @ 0.75 IoU. It also reports the AR (Average Recall). It also reports an overall mAP.
After training my model, I have converted the models to Tensorflow Lite and I have applied post-training quantization on the model. Now I want to calculate similar metrics for my tflite model on my test set.
Can someone guide me on how I can go about doing this? I am trying to understand how I can use the COCOAPI to do this but so far I am not sure how.
A related question is, can someone please tell where I can find documentation for the COCOAPI? I have searched everywhere and I cannot seem to find it.
Thanks