Summary
Prepared a dataset on 2,56,379 seconds of real LIGO noise and Injected simulated gravitational wave signals based on EOB/APR approximations. Performed wave detection parameter estimation of CBC using CNN, ResNet and DenseNet. Pretrained UNet on noise extraction and adapted on wave detection and parameter estimation
Summary
Worked on Cellular segmentation from 7,904 Histopathological tissue images from the PanNuke dataset. Combined Deep Learning architectures with traditional CV approaches to get finer segmentations. Used UNet, SegNet, PSPNet and MaskRCNN for DL based segmentation getting an average IoU of 0.5420. Further refinement with traditional approaches - Region Growing, Mean Shift Segmentation and Spectral clustering improves IoU score of DL models by 20%.
Summary
Developed an AI assistant for research, integrating web search and knowledge retrieval. Used LangChain, Groq API, and Streamlit for real-time AI responses. Integrated Wikipedia, Arxiv, and DuckDuckGo for enhanced insights.
Indian Institute of Technology jodhpur
Marketing
Indian Institute of Technology jodhpur
Publicity and media
DDIT Lab
Recruiting talented individuals, assigning projects based on skills and interests, and effectively managing the team to ensure productivity and collaboration.