3D Infant Pose from RGB Video (Research Internship)

Inria; Montbonnot-Saint-Martin; Remote; EUR 8,900 - 13,000; Full time

Benefits offered by this job

Partial reimbursement of public travel

Possibility of teleworking

Professional equipment (e.g., laptop,

Job summary

The internship at Inria Grenoble Alpes in the Kinetix team focuses on improving the estimation of 3D infant motion from RGB video. You will benchmark state-of-the-art HMR methods on infant videos, create a synthetic dataset with exact 3D ground truth, and fine-tune video-based HMR models with evaluation on real infant footage.

We expect a highly motivated M2 student in CS/Computer Vision, with strong Python skills and experience in deep learning (PyTorch).

Qualifications

• Master's or equivalent required.
• Good programming skills in Python.
• Experience with deep learning frameworks (PyTorch).
• Background in computer vision, 3D geometry, rendering, or human body models (SMPL-type) appreciated.

Responsibilities

• Benchmark state-of-the-art HMR methods on infant videos (image-based and video-based).
• Create a large-scale synthetic video dataset of infants with exact 3D ground truth.
• Run, fine-tune, and improve video-based HMR methods and evaluate gains on real infant videos.

Skills

Python

Deep learning

Computer Vision

3D Geometry

SMPL-type models

Rendering

PyTorch

Education

Master's or equivalent

Tools

PyTorch

Job description

The internship at Inria Grenoble Alpes in the Kinetix team focuses on improving the estimation of 3D infant motion from RGB video. You will benchmark state-of-the-art HMR methods on infant videos, create a synthetic dataset with exact 3D ground truth, and fine-tune video-based HMR models with evaluation on real infant footage.

We expect a highly motivated M2 student in CS/Computer Vision, with strong Python skills and experience in deep learning (PyTorch).