This website is under construction.
Personalized Visual Intelligence (PVI) focuses on developing computer vision algorithms that deliver intelligent services tailored to individual users, ranging from personal item detection and retrieval to customized 3D object capture. By incorporating user-specific knowledge and preferences, PVI enables a wide range of personalized applications that extend beyond the scope of conventional visual perception systems. For example, when integrated with wearable cameras, PVI systems can detect and track personal belongings, helping users locate misplaced items such as keys and wallets in their daily environments. In multi-user settings, such as households, the system can distinguish among family members and accurately recognize and locate each individual's possessions. PVI can also facilitate the digitization of treasured personal objects through customized 3D scanning simply using a smartphone. More broadly, PVI shifts the focus of computer vision from generic object and scene understanding to the perception, modeling, and interaction of user-specific objects and environments. This paradigm opens new opportunities for personalized intelligent services and represents an important step toward human-centered visual intelligence.
This workshop aims to bring together computer vision researchers and practitioners from both academia and industry who are interested in the topics below (but not limited to):
such as text description, reference images, implicit representation, feature representation, and explicit modeling.
including instance detection and tracking, instance reconstruction, instance rendering and editing, etc.
including 3D reconstruction, and interactive modeling.
for the above tasks.
This year, we are excited to announce two challenge competitions:
Instance Detection is an important but fundamental problem in robotics and AR/VR applications. This challenge simulates a use case of indoor robots in terms of locating and recognizing object instances from a cluttered indoor scene at a distance.
Instance 3D capture aims to reconstruct the colored 3D surfaces of objects, facilitating various applications for assistive robots and AR/VR devices. This challenge democratizes this process by utilizing ubiquitous smartphones instead of expensive scanners.
| Time | Session Title | Speaker |
|---|---|---|
| 13:45-14:00 | Opening Remarks | tba |
| 14:00-14:35 | Invited talk #1 | tba |
| 14:35-15:10 | Invited talk #2 | tba |
| 15:10-15:45 | Invited talk #3 | tba |
| 15:45-16:20 | Invited talk #4 | tba |
| 16:20-17:00 | Challenge Summary | Organizers |
Please contact Shu Kong with any questions: aimerykong [at] gmail [dot] com
