Research Focus
Established in 2012, the Rapid-Rich Object Search (ROSE) Lab conducts research in the areas of computer vision, image processing, and pattern recognition. Much of our work involves the application of artificial intelligence (AI) and machine learning approaches, as well as domain generalisation and optimization approaches.
To enable rapid and rich mobile object search with constrained network and computational resources, compact and innovative feature coding, scalable indexing and visual search algorithms are necessary. Topics include object classification, recognition, & retrieval; fashion recommendation. Applications are mainly in the digital advertising, eCommerce, and social media domains.
- Visual Object Search
- Product Search
- Landmark Search
This area of research focuses on the use of traditional as well as machine learning (in particular, deep learning) approaches for video analytics. Topics include anomaly detection, pedestrian detection, person re-identification, object tracking, and action recognition. Applications are mainly in the security and surveillance domains.
- Action Recognition
- Person Re-Identification
- People & Object Tracking
- Visual Anomaly Detection
- Mobile Visual Search
While some parts of our database is confidential, we do have the following datasets that are available for download upon request.
- Warwick-NTU Multi-camera Forecasting (WNMF) database
- NTU CCTV-Fights Dataset
- ROSE-Youtu Face Liveness Detection Dataset
- SIR2 Benchmark Dataset
- NTU RGB+D Action Recognition Dataset
- Video Object Instance Dataset
- Recaptured Images Dataset