Hello,
I am trying to use your trained models to evaluate a dataset of brightfield/phase microscopy images. Before using my own images, I was going to test your model on your dataset of images to make sure it works. I have the LIVECell folder downloaded.
I wanted to evaluate using the following command you provided through terminal:
python train_net.py --config-file --eval-only MODEL.WEIGHTS </path/to/checkpoint_file.pth>
However, I cannot find the "train_net.py" file mentioned anywhere in your GitHub or the LIVECell folder. Is this file present somewhere?
I assumed maybe I should use the "coco_evaluation_resnest.py" as a replacement. I swapped the "COCOEvaluator" class and grabbed this class from "coco_evaluator.py". Would this be a good work-around? Otherwise, what could I do?
Also, I noticed the model.pth, config.yaml, and code script with the evaluation are all present in this terminal command. But, is the set of images ever pointed to? How would I point to the correct images to evaluate?
Thank you anyone for any help.
Hello, I am trying to use your trained models to evaluate a dataset of brightfield/phase microscopy images. Before using my own images, I was going to test your model on your dataset of images to make sure it works. I have the LIVECell folder downloaded. I wanted to evaluate using the following command you provided through terminal: python train_net.py --config-file --eval-only MODEL.WEIGHTS </path/to/checkpoint_file.pth>
However, I cannot find the "train_net.py" file mentioned anywhere in your GitHub or the LIVECell folder. Is this file present somewhere?
I assumed maybe I should use the "coco_evaluation_resnest.py" as a replacement. I swapped the "COCOEvaluator" class and grabbed this class from "coco_evaluator.py". Would this be a good work-around? Otherwise, what could I do?
Also, I noticed the model.pth, config.yaml, and code script with the evaluation are all present in this terminal command. But, is the set of images ever pointed to? How would I point to the correct images to evaluate?
Thank you anyone for any help.