Bengali.AI is a community initiative of machine learning enthusiasts that accelerates language technology research by creating metadata rich datasets and open-sourcing them through competitions. We have recently ended a featured research competition on Kaggle. Link https://www.kaggle.com/c/bengaliai-cv19

As a research assistant at the CCDS lab, I focused on semi-supervised segmentation for ultra-high-resolution images, evaluating context-aware models and contrastive loss-based models with memory banks. I contributed to developing a state-of-the-art model for satellite image segmentation, with plans to evaluate it on a custom satellite dataset from Bangladesh. Additionally, I worked on pelvic bone segmentation from DRR-generated X-rays using CT images, devising a novel data annotation approach… Show more As a research assistant at the CCDS lab, I focused on semi-supervised segmentation for ultra-high-resolution images, evaluating context-aware models and contrastive loss-based models with memory banks. I contributed to developing a state-of-the-art model for satellite image segmentation, with plans to evaluate it on a custom satellite dataset from Bangladesh. Additionally, I worked on pelvic bone segmentation from DRR-generated X-rays using CT images, devising a novel data annotation approach and conducting early benchmarking. Show less

Worked on Project "RadAssist: A Web-based and Artificial Intelligence enabled tele-radiology solution" funded by Information and Communication Technology (ICT) Division of Bangladesh.