Weakly-supervised Semantic Segmentation through Projective Cycle-consistency

Last updated on Nov 28, 2023

The project aims to achieve semantic segmentation using sparse annotations by implementing knowledge transfer mechanisms between multiple 2D images and 3D point clouds through projective cycle consistency. The focus is on leveraging the information from both modalities to enhance the segmentation process, ultimately improving the model’s ability to understand and categorize the visual elements within the data.

Deep Learning Computer Vision Contrastive Learning Multi-modality Semantic Segmentation
Merve Karalı
Merve Karalı
AI Researcher & Data Engineer

Passionate about leveraging technology to drive innovation and solve complex challenges, I am a seasoned professional with expertise in both AI research and data engineering. Holding a Master’s degree in Informatics from Technical University of Munich, my journey has been marked by hands-on experience in crafting and implementing cutting-edge solutions that bridge the realms of artificial intelligence and data-driven insights.