Bhavya Goyal

Bhavya Goyal

Graduate Student
CS Dept. UW-Madison

I am a Grad Student at CS Dept. of UW-Madison.

I am part of Wision Lab where I am advised by Prof. Mohit Gupta. My research revolves around the areas of Computer Vision and Deep Learning. Currently, I am working on Visual Recognition under low-light and motion blur conditions.

Earlier, I worked as a Research Engineer at Visual Understanding Lab of Samsung Research, Seoul for 3 years. Before that, I completed my undergraduate in CS from IIT Delhi where I was advised by Prof. Vinay Riberio


  • Photon Net
    Photon-Starved Scene Inference using Single Photon Cameras
    International Conference on Computer Vision (ICCV), 2021
    Bhavya Goyal, Mohit Gupta
  • ABE
    Attention-based Ensemble for Deep Metric Learning
    European Conference on Computer Vision (ECCV), 2018
    Wonsik Kim, Bhavya Goyal, Kunal Chawla, Jungmin Lee, Keunjoo Kwon


  • University of Wisconsin-Madison
    (Jun'20 - Present)

    Research Assistant
    Wision Lab

    Photon Net
    • SOTA Fine-grained Image Classification under extreme low-light conditions (photons per pixel<0.1).
    • Designed techniques for scene inference tasks like Monocular Depth Estimation on noisy images under photon scarce environments, including architectures with increased shot-noise robustness.

  • Samsung Research, South Korea
    (Sep'16 - Jul'19)

    Research Engineer
    Visual Understanding Lab

    Object Recognition and Retrieval, Smart Refrigerators [Link]
    • Designed and implemented image recognition algorithms for detecting items inside the refrigerator (such as milk, groceries and cans etc.), models used for product recommendation engine in Samsung Smart Refrigerators
    • Used combination of global and attentive deep local descriptors for feature matching, paired with geometric verification of selected keypoints to recognize products that are partially occluded by other items

    Product Search, Bixby Vision [Link]
    • Designed and developed large scale retrieval models for products in online shopping mall images
    • Developed attention mechanism in Deep Neural Networks to ignore background noise in query images, achieves state-of-the-art results on all image major retrieval benchmarks, published in ECCV 2018
    • Optimized recognition models for memory and computational resources for deployment on Samsung phones

  • Headout
    (Jun'16 - Aug'16)

    Full Stack Engineering Intern

    • Designed reliable and scalable payment solution including credit card storage for faster checkouts and fraudulent transactions handling for all transactions using Stripe API integrations. Handles 50,000+ transactions every month
    • Architectured and developed system for discounted pricing of items, pricing localization supporting multiple currencies, coupon management for sellers and inventory update with SpringMVC backend

  • Samsung Electronics, Suwon
    (May'15 - Jul’15)

    Research Intern
    IoT Solutions Lab, S&W Center

    • Implemented indexing algorithms based on B+Tree and μ-Tree in Antelope (DBMS for resource constrained sensor), Reduced flash read/write operations by developing LRU based RAM caching technique for indexing
    • Augmented Antelope DBMS with Thread Safety, DB flushing mechanism, Range based queries & Vacuum operation

  • Zomato, Gurgaon
    (Dec’14 - Jan'15)

    Data Science Intern
    Technology Team

    • Automated spell correction of cuisine names and Indian dishes database for improved search query suggestions
    • Normalizing user ratings of restaurants by detecting spam/fraudulent reviews
    • Developed NodeJS & Redis based server-side app to identify the unnecessary CSS rules. Reduced CSS files by 25%


  • AIMeetsBeauty

    AI Meets Beauty Challenge [ Link ] [ Code ]
    (Jul’18 - Aug'18)

    ACM Multimedia Conference 2018
    • Winner with SOTA results for half million product image recognition
    • Developed CNN based retrieval model using attention module to ignore background clutter in product images, approx nearest neighbor search for product images in large scale retrieval DB
  • Tiger Re-ID

    Tiger ReID in Wild [ Code ] [ Slides ] [ Paper ]
    (Jun’19 - Jul'19)

    Prof. Yin Li

    • Proposed architecture using object detection and re-id which encourages diversity among feature embeddings to get more discriminative features which boosts the performance on most retrieval benchmarks.
  • Stack Exchange

    Stack-Exchange Tag Prediction [ Code ] [ Slides ] [ Paper ]
    (April’16 - May'16)

    Prof. Mausam

    • State-of-the-art results for predicting tags/labels for questions on different Stack Exchange portals
    • Classification of meta features from text and code snippets in questions using SVM
    • KMeans to cluster word embeddings from Google's pre-trained Word2Vec model and ensembled with model trained with term affinity of tags and words
  • Next Generation Wifi [ Code ] [ Report ] [ Slides ]
    (Aug’15 - Nov'15)

    Prof. Vinay Ribeiro

    • Dynamically assigning primary and secondary channels to access points for best performance in a given settings
    • Improving throughput in dense deployment scenarios through central controller or Wi-Fi protocols
    • Analysing various proposals in 802.11ax and proposing model for adaptive clear channel assessment
  • Glucosensor [ Code ]
    (Jun’14 - Jul'14)

    Prof. Sandeep K Jha

    • Designed and implemented low cost solution to determine glucose level from RGB images of Diabetes test strip captured by smartphone, an alternative to glucose meter (costs around $50), for CBME, IIT Delhi
    • Estimates color change of pH indicator with high robustness to illumination and camera angles using multiple normalization techniques
  • Quoridor AI Bot [ Code ]

    Prof. Mausam

    • Designed an adaptive AI player for finite time Quoridor N-M-K tournament for three configurations
    • Engineered the game as an instance of minimax search with alpha beta pruning using evaluation functions