Praneet C. Bala

Praneet C. Bala

I build systems that see, reconstruct, and reason about the 3D world.

Resume

I’m a Ph.D. candidate in Computer Science at the University of Minnesota, Twin Cities, advised by Dr. Hyun Soo Park and Dr. Jan Zimmermann. My research focuses on 3D Computer Vision, self-supervised learning, and generative modeling, building systems that perceive, reconstruct, and predict the physical world from visual data.

My work spans the full stack: multi-camera capture systems, contrastive representation learning, large-scale dataset creation, latent diffusion models, and neural rendering. I have 4 publications in Nature Communications, IJCV, and eLife, and a paper on domain-aware 3D motion forecasting under review at ECCV 2026.

My research maps directly onto problems in autonomous vehicles, embodied AI, and generative 3D.

Projects

3D body motion forecasting teaser
Learning to Forecast Domain-Aware 3D Body Motion
Self-supervised framework that jointly learns 3D motion reconstruction and future motion generation from 2D domain videos. Uses a spatiotemporal transformer (PoseFormer) for 2D-to-3D lifting and a latent diffusion model (BeLFusion) for long-term forecasting. Achieves competitive ADE/FDE on Human3.6M and outperforms baselines on cross-domain MPI-INF-3DHP without 3D domain supervision.
PyTorch Spatiotemporal Transformer Latent Diffusion Self-supervised Domain Adaptation
European Conference on Computer Vision, 2026 [Under Review]
Secondary landmark detection teaser
Self-supervised Secondary Landmark Detection
Self-supervised method using 3D representation learning and multiview contrastive learning to detect anatomically consistent secondary landmarks without manual annotation. Uses Procrustes-normalized triangulation and twin convolutional pose machines to enforce geometric and semantic consistency across views. Achieves PCKh@0.5 of 0.75 on OpenMonkeyPose, validated across macaques, humans, and flies with less than 10% labeled data.
TensorFlow Contrastive Learning Multi-view Geometry 3D Representation Learning Pose Estimation
International Journal of Computer Vision, 2023
OpenMonkeyStudio teaser
OpenMonkeyStudio
Engineered a 62-camera markerless motion capture system for high-fidelity 3D body tracking of freely moving macaques in unconstrained environments. Built an end-to-end pipeline spanning multi-view calibration, 2D pose estimation, and 3D triangulation. Released a public dataset of 195,228 annotated pose frames across 13 body landmarks and diverse activities.
TensorFlow MATLAB Multi-view Geometry 3D Reconstruction Pose Estimation
Nature Communications, 2020
View all projects →

Publications

Bala et al., “Learning to Forecast Domain Aware 3D Body Motion by Watching Single View Videos.” European Conference on Computer Vision, 2026. Under Review

News

Service

Reviewing
Invited Talks
Automated Markerless Pose Estimation in Freely Moving Macaques with OpenMonkeyStudio
UMN Center for Neuroengineering seminar series
UMN Visual Computing & AI seminar series
Teaching