Snowmass Mountain, Colorado
Snowmass, CO · March 2020
Siddhartha Gairola

Siddhartha Gairola

PhD student in computer vision & machine learning · MPI for Informatics & ISTA

I am an ELLIS PhD student at the Max Planck Institute for Informatics and ISTA, advised by Bernt Schiele and Francesco Locatello. My research centres on representation learning for vision and vision-language foundation models. I am especially interested in how these models can be used for visual understanding tasks such as detection, segmentation and localisation, and in making deep networks interpretable.

Before the PhD I was a Research Fellow at Microsoft Research India, building low-cost smartphone diagnostics for eye care, and a research intern at Adobe. I did my MS and B.Tech at IIIT Hyderabad with P. J. Narayanan.

When not working on my research, I like to play the piano 🎹 and guitar 🎸, listen to music 🎧, read non-fiction 📚, drive motorcycles 🏍️, and go for a run 🏃 or hike 🥾. I am also really fascinated by paradoxes ♾️ (you can find some here), and I wish I had Hermione's Time-Turner ⏳ to do much more in a day. See more at Personal.

News

Older news
  • WACVSmartKC++ accepted at WACV 2025.
  • ICLRHow to Probe accepted at ICLR 2025.
  • Invited talk at the GMUM Workshop, Jagiellonian University on XAI in visual models.
  • Started my PhD at MPI for Informatics as part of the ELLIS PhD program.
  • Auto-retinoscopy and keratoconus classifier papers published at IMWUT and EMBC.
  • SmartKC published at IMWUT; RespireNet at EMBC 2021.

Selected workall 14 papers ↓

NeurIPS 2026

Do vision models still recognise an object once its usual neighbours are gone? A dataset and study of context reliance.

arxiv · code · dataset · bibtex (coming soon)
ICML 2026 Spotlight

Distribution-aware attribution for ViTs: stable, high-resolution saliency maps without patch-grid artifacts.

CVPR 2026

One label-free alignment step turns DINOv3 or SigLIP2 into an inherently interpretable B-cos model.

abstract · arxiv · code · project page · bibtex

PublicationsGoogle Scholar ↗

Show
2026
  • DisParQ: Self-Supervised Part Concepts for Interpretable Vision Foundation Models
    PreprintUnder submission
    arxiv · code · bibtex (coming soon)
  • Stranger Things: When Objects Appear Without Their Typical Neighbours
    NeurIPSConference on Neural Information Processing Systems (NeurIPS)
    arxiv · code · dataset · bibtex (coming soon)
  • DAVE: Distribution-aware Attribution via ViT Gradient Decomposition
    ICML SpotlightInternational Conference on Machine Learning (ICML)
  • Align Once to Explain: Feature Alignment for Scalable B-cosification of Foundational Vision Transformers
    Raphael Maser*, Siddhartha Gairola*, Sukrut Rao, Bernt Schiele
    CVPRIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
    abstract · arxiv · code · project page · bibtex

    Spotlight at the HOW Workshop and poster at the XAI4CV Workshop, CVPR 2026.

2025
  • How to Probe: Simple Yet Effective Techniques for Improving Post-hoc Explanations
    ICLRInternational Conference on Learning Representations (ICLR)
  • SmartKC++: Improving Performance of Smartphone-Based Corneal Topographers
    Vaibhav Ganatra, Siddhartha Gairola, Pallavi Joshi, Anand Balasubramaniam, Kaushik Murali, Anusha Varadharajan, Bhavana Mallikarjuna, Nipun Kwatra, Mohit Jain
    WACVIEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
2022
  • Keratoconus Classifier for Smartphone-based Corneal Topographer
    Siddhartha Gairola, Pallavi Joshi, Anand Balasubramaniam, Kaushik Murali, Nipun Kwatra, Mohit Jain
    EMBCIEEE Engineering in Medicine and Biology Conference (EMBC)
  • Towards Automating Retinoscopy for Refractive Error Diagnosis
    Aditya Aggarwal, Siddhartha Gairola, Uddeshya Upadhyay, Akshay P. Vasishta, Diwakar Rao, Aditya Goyal, Kaushik Murali, Nipun Kwatra, Mohit Jain
    IMWUTProceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 6(3)
2021
  • SmartKC: Smartphone-based Corneal Topographer for Keratoconus Detection
    Siddhartha Gairola, Murtuza Bohra, Nadeem Shaheer, Navya Jayaprakash, Pallavi Joshi, Anand Balasubramaniam, Kaushik Murali, Nipun Kwatra, Mohit Jain
    IMWUTProceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 5(4)
  • RespireNet: A Deep Neural Network for Accurately Detecting Abnormal Lung Sounds in Limited Data Setting
    Siddhartha Gairola, Francis Tom, Nipun Kwatra, Mohit Jain
    EMBCIEEE Engineering in Medicine and Biology Conference (EMBC)
    arxiv · pdf · code · bibtex
2020
  • SimPropNet: Improved Similarity Propagation for Few-Shot Image Segmentation
    Siddhartha Gairola, Ayush Chopra, Mayur Hemani, Balaji K.
    IJCAIInternational Joint Conference on Artificial Intelligence (IJCAI)
  • Unsupervised Image Style Embeddings for Retrieval and Recognition Tasks
    Siddhartha Gairola, Rajvi Shah, P. J. Narayanan
    WACVIEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
2018
  • Identifying Clickbait: A Multi-Strategy Approach Using Neural Networks
    Vaibhav Kumar*, Dhruv Khattar*, Siddhartha Gairola*, Yash Kumar Lal*, Vasudeva Varma
    SIGIRACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)
    arxiv · acm · code · bibtex
  • Find Me a Sky: A Data-Driven Method for Color-Consistent Sky Search and Replacement
    Saumya Rawat*, Siddhartha Gairola*, Rajvi Shah, P. J. Narayanan
    MMMInternational Conference on Multimedia Modeling (MMM)
Thesis
  • Image Representations for Style Retrieval, Recognition and Background Replacement Tasks
    Siddhartha Gairola
    MS ThesisMaster's Thesis, IIIT Hyderabad, 2020

*equal contribution · click bibtex on an entry to copy or download its citation

Backgroundfull CV (PDF) ↗

Experience

Education

Academic service & more

Reviewing

ICML 2023–26NeurIPS 2022–23, 2025ICLR 2022–24CVPR 2024–26ECCV 2024ICCV 2025IHCI 2021

gold reviewer, ICML 2026

Talks4 talks

  • Intriguing Applications and Overlooked Pitfalls of XAI in Visual ModelsGMUM Workshop, Jagiellonian University · invited talk · slides
    Oct 2024
  • RespireNet: Detecting Abnormal Lung Sounds in Limited Data SettingsEMBC 2021 · paper talk · video
    2021
  • SimPropNet: Improved Similarity Propagation for Few-shot SegmentationIJCAI 2020 · paper talk · video
    2020
  • Unsupervised Image Style Embeddings for Retrieval and Recognition TasksWACV 2020 · paper talk · video
    2020

Teachingteaching assistant · 6 courses

Open source

Volunteering

Writing & resources

Guides I maintain

Templates & checklists

Elsewhere

  • Mediumoccasional posts on research, general thoughts, and personal things
  • Personal pagesome personal, impersonal, and random stuff