Sai Kumar Reddy Manne

Research

• 
  Overcoming Small Data Limitations in Video-Based Infant Respiration Estimation

  L. Song, H. Bishnoi, Sai Kumar Reddy Manne, S. Ostadabbas, B.J. Taylor, & M. Wan

  IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026

  paper code

  AIR-400: a 400-video annotated dataset and the first reproducible pipelines for contactless infant respiration estimation using spatiotemporal neural processing enhanced by optical flow.
• 
  Classification of Infant Sleep-Wake States from Natural Overnight In-Crib Sleep Videos

  S. Moezzi, M. Wan, Sai Kumar Reddy Manne, A. Mathew, S. Zhu, B. Galoaa, ... & S. Ostadabbas

  IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025

  paper code

  A video-based system for classifying infant sleep vs. wake states from natural overnight in-crib recordings without wearable sensors.
• 
  NOISe: Nuclei-Aware Osteoclast Instance Segmentation for Mouse-to-Human Domain Transfer

  Sai Kumar Reddy Manne, B. Martin, T. Roy, R. Neilson, R. Peters, M. Chillara, C.W. Lary, K.J. Motyl, & M. Wan

  IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPR-W), 2024

  paper code

  A nuclei-aware instance segmentation model for osteoclast identification, using a two-stage training strategy to generalize from mouse to human bone samples.
• 
  WB LUTs: Contrastive Learning for White Balancing Lookup Tables

  Sai Kumar Reddy Manne & M. Wan

  arXiv preprint, 2024

  arXiv code

  A contrastive learning framework that generates 3D lookup tables for efficient and accurate white balance correction on camera ISP pipelines.
• 
  Subtle Signals: Video-based Detection of Infant Non-nutritive Sucking as a Neurodevelopmental Cue

  S. Zhu, M. Wan, Sai Kumar Reddy Manne, E. Zimmerman, & S. Ostadabbas

  Computer Vision and Image Understanding, 2024

  arXiv

  A video-based method for detecting non-nutritive sucking in infants — an early neurodevelopmental biomarker — without contact sensors.
• 
  Automatic Infant Respiration Estimation from Video: A Deep Flow-Based Algorithm and a Novel Public Benchmark

  Sai Kumar Reddy Manne, S. Zhu, S. Ostadabbas, & M. Wan

  International Workshop on Preterm, Perinatal and Pediatric Image Analysis (MICCAI-W), 2023

  arXiv code

  AIRFlowNet: a deep optical flow-based model with spectral bandpass loss, plus AIR-125 — achieving 2.9 bpm MAE for contactless infant respiration estimation.
• 
  TUPAC: Thermal Image Upsampling with Pixel Adaptive Convolutions

  Sai Kumar Reddy Manne

  Thermal Image Super-Resolution Challenge, PBVS, CVPR 2023

  challenge code

  Guided thermal image super-resolution using pixel adaptive convolutions for the PBVS thermal SR challenge at CVPR 2023.
• 
  Moiré Removal CNN for Smartphones

  Sai Kumar Reddy Manne · Samsung R&D

  Commercialized in Samsung Galaxy S22, 2021

  media patent

  First-ever moiré removal model for smartphones. Optimized for multiple accelerators, improving user MOS from 2.6 to 4.1. Shipped in Galaxy S22.
• 
  Old Photo Enhancer for Gallery Images

  Sai Kumar Reddy Manne · Samsung R&D

  Commercialized in Samsung Galaxy S21, 2020

  media

  Deep neural network for enhancing old and degraded gallery photos, supporting 1 MP to 108 MP resolutions. Shipped in Galaxy S21.
• 
  Asymmetric Wide Tele Camera Fusion for High Fidelity Digital Zoom

  Sai Kumar Reddy Manne, B.P. Prasad, & K.G. Rosh

  International Conference on Computer Vision and Image Processing, Springer, 2019

  paper

  Multi-camera fusion algorithm combining wide and telephoto inputs for high-fidelity digital zoom, improving PSNR by 20%.

Academic Service