Cross-Domain Instance Segmentation & Style Transfer for Industrial Parts Inspection скачать в хорошем качестве

Cross-Domain Instance Segmentation & Style Transfer for Industrial Parts Inspection

This video is a demonstration of my project on Instance Segmentation and Domain Adaptation for industrial defect detection. This project focuses on improving mode performances without retraining by using CycleGAN for domain shifting. This demo covers the basic idea, need of the project, dataset used, model training and evaluation, domain adaptation and qualitative results in industrial scenarios. Here is my github link to all the codes demonstrated in the video https://github.com/AamenaMahudhawala/... Here is the link to the original Neu-Seg Dataset https://universe.roboflow.com/school-... Here is the link to the Cyclegan repo i used in the video https://github.com/junyanz/pytorch-Cy... This project was developed as an academic / research-oriented project.