How to measure accuracy, loss, F1-score,and more on jetson nano with jetson-inference?
And how to show the confusion matrix?
Hi @m012110024, while training an image classifier or object detector with PyTorch using the train.py or train_ssd.py scripts, they will automatically output the loss and train.py will automatically output the accuracy. With train_ssd.py, you can use the --validate-mean-ap flag so that it computes the overall and per-class Mean Average Precision (mAP) at the end of each epoch.
To compute the F1-score, confusion matrix, ect you could use a library like torchmetrics or scikit-learn and integrate them into the training scripts. I haven’t done this so you would need to make the necessary modifications.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.