Philips, Internship: Synthetic Data Generation for Dental AI

This internship develops a scalable pipeline for generating photorealistic synthetic intraoral images and automatic ground-truth labels from 3D jaws. Tooth detection and instance segmentation are the primary validation case. The same approach should also support other intraoral AI algorithms, such as gumline localization and detection of cavities or other oral conditions. Building on previous work in 3D jaw reconstruction and SAM mask generation/distillation, the project will quantify how synthetic data affects accuracy, robustness and the synthetic-to-real domain gap.

Your role:

• Review 3D/synthetic dental data, rendering and domain-randomization methods.
• Build a data-generation pipeline from reconstructed or parametric 3D jaws.
• Vary anatomy, camera pose, lighting, materials and oral conditions.
• Export images with masks, depth, normals and tooth labels.
• Train/evaluate tooth detection and identification models; report conclusions.
• Draft a technical note or paper; present results at a Philips R&D forum.

• MSc. in Data Science, CS, AI, Mathematics, Physics or Biomedical Engineering.
• Experience with Python programming is essential.
• Good knowledge of deep learning and computer vision in PyTorch.
• Passion for 3D modelling, rendering and synthetic data generation.
• Experience with Blender, 3D reconstruction or generative AI is a plus.
• Language: good written and spoken English.
• Previous internship experience is not necessary but is a plus.
• Good communication skills and enthusiasm to work in an international team