• Feature-Driven Neural Network Approach for Accurate Rice Grain Classification
  Krishnam Soni, Sanjay Kumar Tehariya
  CVR 2025 (Accepted)
  
  • Paper accepted at the 9th CMVIT conference, but could not go due to financial constraints.
  • Developed a ML model to classify rice grains using 4,500 features per image, achieving high performance with a lightweight multilayer perceptron (MLP). Identified key features through systematic evaluation.
• Segmenting Optical Coherence Tomography (OCT) Scans Using Multi-Scale Modified UNet Architecture
  Krishnam Soni, Karri Karthik, Manjunatha Mahadevappa
  IEEE INDICON 2024 (Presented)
  
  • Developed an UNet-based deep learning model for OCT retinal segmentation, improving accuracy over traditional methods. Achieved performance close to intra-observer results on the AROI dataset.