I have been training an object detection model with tensorflow object detection api. I have questions regarding the ways to evaluate the quality of the model.
On tensorboard, I can display the loss accross the steps for the train dataset but not for the eval dataset. Also I can only check the accuracy (mAP) at the very final stage of the training process. My question is: is there a way to visualize the evolution of accuracy and loss for both the train and the eval datasets while training?
I have tried different solution: use model_main.py instead of model_main_tf2.py, run eval.py and train.py as suggested below but it gave me errors.
Tensorflow Object Detection API only shows loss value. How to get mAP value?
It is possible these options are merely not available yet with tf2 object detection api:
How to train and eval at the same time in Tensorflow Object Detection API v2
model = ssd_mobilenet_v2_fpnlite_320x320_coco17_tpu-8
python = 3.8.5
tensorflow = 2.4.1
