Generative AI for Text-Based Image Retrieval System
When investigating cases, police officers often spend a significant amount of time reviewing surveillance footage to identify key information such as suspect characteristics, vehicles, street names, and license plates. If an AI system could automatically analyze specific objects and generate reports on their appearance times, and provide a text-based Q&A function to quickly obtain necessary information, it would greatly reduce the time officers spend watching videos, decrease visual fatigue, and enhance case efficiency.
The Kaohsiung City Government Police Bureau participated in the "AI Application Competition" hosted by the Digital Development Department's Digital Industry Bureau in 2024. After technical review and matchmaking, the project collaborated with the IVC Laboratory of Tainan University to develop a "Text-Based Image Search System Using Generative AI," aiming to improve crime investigation efficiency.
System Functions and Advantages
TheGenerative AI for Text-Based Image Retrieval System significantly reduces the time officers spend viewing and analyzing footage, enhancing investigation efficiency. Its main functions include:
Automated Video Analysis: Users can input the names of objects they want to detect, such as yellow cars, license plates, street names, or convenience stores. The system analyzes the video content and generates a report indicating when these objects appear.
Integration with ChatGPT Technology: Users can quickly obtain video content through a Q&A interface, making operations convenient and intuitive.
Electronic Fence Function: Allows setting specific analysis ranges to reduce unnecessary object interference, further improving analysis accuracy.
This system not only effectively reduces labor costs and visual fatigue but also significantly shortens case investigation times, marking a significant breakthrough in policing technology.
Future Outlook
The Generative AI for Text-Based Image Retrieval System showcases the technological advancement and modernization of police tools, enabling rapid case resolution and enhancing public perception and trust in police work. In the future, this system can be referenced by neighboring city police departments and promoted nationwide to comprehensively improve policing efficiency.
System Interface
