Research

Research Interests

My research interests include Computer Vision, ML Robustness, Diffusion Models, and Efficient Deep Learning.

Ph.D. Focus

Real-world images frequently have different corruptions/perturbations, like motion blur, noise (caused by low-light conditions), and digital compression, appearing in various digital versions of images/videos. Specifically, computer vision tasks face challenges with image degradation, leading to diminishing model performance or reliability. My Ph.D. focuses on narrowing the performance gap across diverse computer vision tasks, bridging the disparity between real-world and pristine/clean images.

Publications

  • Dinesh Daultani, Hugo Larochelle, “Consolidating separate degradations model via weights fusion and distillation,” Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops, 2024. [Paper], [Code]
  • Dinesh Daultani, Masayuki Tanaka, Masatoshi Okutomi, Kazuki Endo, “ILIAC: Efficient classification of degraded images using knowledge distillation with cutout data augmentation,” Proceedings of Image Processing: Algorithms and Systems Conference, at IS&T Electronic Imaging 2023. [Paper]
  • Image classification of degraded images using knowledge distillation and robust data augmentation. (Submitted in a journal)

Patents

  • Ensembles optimization using genetic algorithms: Dinesh Daultani, Bruno Charron, “Information processing apparatus, method for information processing and program thereof”, PCT JP2020/001767, filed on 2020/01/20, granted on 2020/11/27. [Link]
  • Anomaly detection in cybersecurity: Dinesh Daultani, Bruno Charron, “Anomaly detection using user probability”, PCT JP2020/037004, filed on 2020/09/29, granted on 2022/06/01. [Link]
  • Lifelong learning: Dinesh Daultani, Mitsuru Nakazawa, “Information processing device, information processing method, and program”, PCT/JP2021/044858, filed on 2021/12/07, granted on 2023/06/15. [Link]
  • Hierarchical image classification: Dinesh Daultani, Mitsuru Nakazawa, “Information processing device, information processing method, program, and learning model”, PCT/JP2021/044857, filed on 2021/12/07, granted on 2023/06/15. [Link]
  • Two other patents on topics related to azimuth angle measurement/screw inspection using drone images.