Mohit Kumar, Vijayarajan Rajangam, Alex Noel Joseph Raj, Sangeetha Nagarajan, Parthasarathy Ramanujam
Scientific Reports (Nature Portfolio) · Published online — Sep 2026
A hybrid GAN framework (HINT-Net) for structure-preserving intensity normalization of medical images across multiple intensity domains, using a single unified generator with an attention-based architecture and a hybrid contrastive loss (a Gated Multi-Domain CUT design), achieving an SSIM of 0.9858 and outperforming CycleGAN, StarGAN, and CUT.
V. Ananthakrishnan, Sounak Shome, Mohit Kumar, Suryakiran R, Dr. Tharaga Sharmilan, Dr. Gnana Swathika O. V
IEEE i-PACT 2025 · December 2025
A lightweight MobileNet-based disease classification system integrated with AWS S3 for automated tomato plant disease detection, achieving >90% accuracy with <150ms inference latency on mobile devices.
Mohit Kumar, Diya Ravishankar, Yuganshu Kumar, Sudhakaran G
2026 Intl. Conference on Computational Intelligence and Network Systems (CINS 2026) · Accepted · 29–30 Oct 2026 — BITS Pilani Dubai Campus, UAE
A system blending neural models with rule-based reasoning to determine legal aid eligibility, achieving F1-score of 0.985.
Diya Ravishankar, Mohit Kumar, Yuganshu Kumar, Vijayarajan Rajangam, Sangeetha Nagarajan
Under Review
Novel end-to-end framework for layout-free parking vacancy detection achieving mAP50 of 98.99%.
Dr. Balasundaram A, Diya Ravishankar, Mohit Kumar, Aditi Bhargav
Under Review
Comparative study achieving 99.99% accuracy with HAR-GCNN on PAMAP2 dataset.