Vacancies
| Details | |
|---|---|
| Project Title | Privacy-Preserving Visual Recognition via Foundation Model Embeddings and Self-Supervise Latent Contrast Learning |
| Grant | UTARRF 2025 Cycle 2 |
| Campus | Sg Long campus |
| Stipend | RM2,000.00 per month |
| Entry Requirement | 1. Bachelor’s degree in Science, Computing, Mathematics, Computer Science, or a related field 2. Eligible to enroll in a Phd’s program 3. Proficient in English (both written and spoken) |
| Job Description | This research aims to build visual recognition systems that work even when labels are scarce and raw images can’t be shared for privacy reasons. Using modern Vision Foundation Models like SAM and CLIP, the system automatically extracts meaningful representations from unlabeled images. These are then improved with self-supervised contrastive learning so the model can learn strong features without accessing sensitive data.
The result is a privacy-preserving, data-efficient framework suited for real-world areas such as healthcare, biometrics, and industrial inspection.
Objectives:
Create a privacy-friendly auto-annotation pipeline using SAM and CLIP.
Improve embeddings with self-supervised contrastive learning (via SLC) for robust performance with minimal labeled data. |
| Field of Study Related to the Research Project | Doctor of Philosophy (Computer Science) |
| How to Apply | If you are interested, please contact Dr. Lai Yen Lung via email at laiyl |
