【BullyProof】拒絕覇凌 – Bullying Detection using Computer Vision

參賽團隊名稱：BullyProof

欺凌通常發生在無人監管的時期，當沒有當局在場時，學生更有可能進行身體虐待或攻擊。本案運用AI電腦視覺系統來進行監視預防，該系統包含一個動作檢測層，它保存潛在暴力動作的視頻，並將對所述視頻的引用存儲在 CSV 文件中。 所述文件作為一種記錄，可以很容易地訪問和審查，以評估一個行為是否真的是暴力行為。

提案動機

There are several reasons why we undertook this project: the primary reason is to create a safe and conducive learning environment for students. It is society’s responsibility to provide such an environment, and ensuring the physical safety of students is paramount to doing so. Students who are bullied often lose interest in attending school, and their learning is negatively affected. The second reason for the creation of this project is the need for constant supervision. Bullying typically occurs during unsupervised periods, and students are more likely to engage in physical abuse or aggression when there are no authorities present.

Lastly, there is a need for automation in this area. With the increasing number of CCTV cameras in learning institutions, it is not possible for humans to supervise the footage in real-time. Additionally, reviewing recorded footage is a tedious task. It is much more efficient to only review the footage that has been previously flagged as violent. For all these reasons, we decided to implement a detection system for CCTVs with the goal of both identifying and preventing this kind of undesired behavior.

成果介紹

The result of this project will be a system that can detect and record potentially violent actions in video footage. The system will use computer vision algorithms to analyze video in real-time and identify actions that may be considered violent. When a violent action is detected, the system will save a video of the action and store a reference to the video in a CSV file. This file will function as a record that can be easily accessed and reviewed by authorized personnel to assess the severity of the detected action.

The system will be trained using a large dataset of labeled video footage and will be able to adapt and improve over time. Once deployed, the system will be able to monitor a public area and alert security personnel in real-time if violent action is detected. This will provide a valuable tool for preventing violence and ensuring public safety.