Greetings.

Abhishek Kuriyal

Researcher in Geospatial Artificial Intelligence and 3D Computer Vision.

👋 Hello!

I am a recent Master’s graduate in Data Science and Artificial Intelligence from IISER-B. My prior research focuses on semantic segmentation, interpretability of point cloud architectures, and spatial generalization in satellite imagery.

Recently, I worked as a Research Engineer at Imagine Lab, École des Ponts ParisTech, France, and previously interned at CRDIG, Université Laval, Canada, as a summer intern.

I am currently seeking Researcher/PhD opportunities in 3D computer vision, mathematical modeling and related fields.

I am also a 3D artist! Would love to work on simplifying the complex Geometry Node workflows using vision driven intelligence.

📚 Research Work

CoDEx

CoDEx: Combining Domain Expertise for Spatial Generalization in Satellite Image Analysis

Abhishek Kuriyal, Elliot Vincent, Mathieu Aubry, Loic Landrieu

EarthVision 2025 - CVPR Workshop

Spatial domain generalization for satellite imagery using region-specific expert models.

pCTFusion

pCTFusion: Point Convolution-Transformer Fusion with Semantic Aware Loss

Abhishek Kuriyal, V. Kumar, B. Lohani

Journal of Computer Science (Springer Nature), 2024

A hybrid convolution-attention architecture emphasizing semantic boundaries with position-aware loss.

pGS-CAM

pGS-CAM: Interpretable LiDAR Point Cloud Segmentation via Gradient-Based Localization

Abhishek Kuriyal, V. Kumar

ICLR 2023 (Tiny Papers Track)

First study on gradient-based localization for LiDAR semantic segmentation. Enables interpretable saliency maps in 3D networks.

Tree Extraction

Individual Tree Extraction in Airborne LiDAR Point Clouds of Boreal Forest Environments

Abhishek Kuriyal

Master Thesis

Deep learning-based individual tree detection from airborne LiDAR in Canadian boreal forests.

🏆 Awards & Fellowships

  • European Research Council (ERC): Project DISCOVER, number 101076028
  • MITACS GRI: Awarded MITACS GRI fellowship (2023) for internship in Canada among students participating over 17 countries globally.
  • TiHAN IIT-H Startup: Selected for initial startup funding of 7200 USD by TiHAN IIT-Hyderabad for a project related to Conditional LiDAR generation.
  • Academic Poster Presentation: Awarded best academic poster presentation on National Engineer’s Day (15 September 2022), India held at IISERB campus.

📖 Expository Articles

DDPMs

Unpacking DDPMs: What's Really Going On?

Abhishek Kuriyal

Technical Note, 2026

Diving deeper into the mathematical beauty and intricacy of Denoising Diffusion Probabilistic Models (DDPMs).