How-to-Fine-Tune-Object-Detection-Model-on-ImageJ-ROI-Custom-Dataset-in-PyTorch

Background: Deep learning models such as YOLO and Faster RCNN are generally used for object detection tasks. Annotation files are usually fixed-format files such as XML files that use specialised software to annotate images. ImageJ is a powerful open-source image processing software that is widely used in many fields, especially in scientific research and industry. Common image processing tasks in ImageJ also include marking ROI (region of interest) for subsequent processing.

Aim: Many tutorials for fine-tuning object detection models with custom data are based on common annotation formats such as XML files. This tutorial shows how to directly read ImageJ's ROI file and generate model input, how to train, and evaluate test results.

Data: The dataset used for this tutorial comes from a public dataset: https://www.kaggle.com/datasets/andrewmvd/face-mask-detection
I converted the original XML format annotation file to ImageJ's ROI file format for tutorial explanation.

The input image needs to be in RGB mode, with a bit depth of 8 for each color channel.

Environment：The tutorial code file is mainly run on Google Colab.

lyucong7/How-to-Fine-Tune-Object-Detection-Model-on-ImageJ-ROI-Custom-Dataset-in-PyTorch

ErrorLooks like something went wrong!

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Languages

Error
Looks like something went wrong!

Packages