shradhautk Goto Github PK

ai-microscopy-workshop

Python code files to be used in AI MICROSCOPY WORKSHOP AT UTK

comparing-five-loss-functions-iou-dice-bce-wbce-distance

By utilizing this repository, individuals can determine the most effective loss function for accurately segmenting transmission electron microscopy images that contain material defects .

coupling-powerful-lossfunction-bcewithiou

Uniting Binary Cross-Entropy with IOU for Improved Performance

creating-regionproposalnetwork-applying-nonmaxsuppression

Code outlines RPN creation with NMS integration for Feature Pyramid Network (FPN). Functions manipulate bounding boxes and a custom layer processes RPN-generated proposals. Finally, custom Keras layer generates target data for object detection training.

dislocation-loop-analysis-through-generative-ai-leveraging-variational-autoencoders

handling--imbalanceddata-wbce-bce

Discover how WBCE, intended for data balance, fell short in our performance evaluation. Explore insights and code for imbalanced datasets, based on Helium Bubble Detection research

he-bubble-video-tracking

This project utilizes advanced computer vision techniques, including Mask R-CNN, to accurately identify and monitor helium-induced bubbles in videos obtained from TEM. Researchers and developers can leverage this resource to enhance material characterization under extreme conditions, ensuring safety and reliability in nuclear environments.

lossfunc-comparison

occlusion-challenge-in-early-yolo-versions-object-detection-issues

Explore occlusion challenges faced by early YOLO (You Only Look Once) versions V!, V2, V3, focusing on object detection issues.

python_basics

This repository contains Python basics practice

transfer-learning-with-the-maskrcnn-model-pre-trained-on-the-coco-dataset

This code example demonstrates the use of transfer learning with the Mask R-CNN model pre-trained on the COCO dataset to perform object detection on a random image. Here's a step-by-step breakdown

use-of-erosion-dilation-technique

nvestigating the Impact of Post-processing Techniques like Erosion and Dilation on the Performance of Models Utilizing Various Loss Functions, including Binary Cross-Entropy, Weighted Binary Cross-Entropy, Distance, Dice, and IOU.

yolo-versions-1to7-comparison-for-helium-induced-bubble-detection-in-tem-micrographs

This repository can be used to enhance the accuracy and robustness of object detection models, especially when dealing with scenarios involving occluded objects. Researchers and developers interested in improving YOLO-based object detection algorithms can reference and utilize the code and functions provided here to address occlusion challenges.