SeekNspot - Missing person Detection

Inspiration

While understanding how missing person cases are handled in real life, we noticed that a major part of the investigation still involves manually going through CCTV footage. Officers often have to sit for hours, watching videos frame by frame, trying to identify a single individual. This process is not only slow but also mentally exhausting, and in situations where time is critical, such delays can reduce the chances of finding someone quickly.

This made us question why such an important task is still dependent on manual effort when computer vision has advanced so much. We were inspired to build a system that could reduce this effort and assist in speeding up the identification process using AI.

What it does

Our project automates the process of identifying missing individuals from video footage. Instead of manually scanning videos, the system allows users to upload a recorded video, which is then processed automatically.

The system extracts frames from the video, detects faces in those frames, and compares them with a database of missing persons. If a potential match is found, it is highlighted along with a confidence score. This helps reduce hours of manual work into a much faster and more efficient process, making it easier to analyze large amounts of video data.

How we built it

We built the system by combining computer vision techniques with deep learning-based face recognition. The process begins with detecting faces in each frame of the video. Once detected, each face is converted into a numerical representation, known as an embedding, which captures its unique features.

These embeddings are then compared with stored embeddings of missing individuals using similarity measures. For the interface and deployment, we used Streamlit to create a simple and interactive application that allows users to upload videos and view results easily.

Initially, we aimed to implement real-time video processing. However, due to deployment limitations and the heavy nature of the models, we adapted the system to work with pre-recorded videos. This made the application more stable and ensured that it could run smoothly in a cloud-based environment.

Challenges we ran into

One of the main challenges we faced was handling real-time video processing in a deployable environment. While it was possible during development, it became difficult to maintain performance and stability when trying to deploy it due to resource constraints and large dependencies.

We also faced issues with processing speed, as analyzing every frame of a video increased latency. Another challenge was maintaining accuracy while minimizing false positives, which required careful tuning of similarity thresholds and processing logic. These challenges pushed us to rethink our approach and focus on building a more practical and reliable solution.

Accomplishments that we're proud of

We are proud of successfully turning a real-world problem into a working solution. Automating a process that is traditionally manual and time-intensive was a significant achievement for us. We also managed to adapt our system to work within real constraints, making it deployable and easy to demonstrate without losing its core functionality.

What we learned

This project helped us understand that building an AI model is only one part of solving a problem. Making that model usable, efficient, and deployable in real-world scenarios is equally important. We learned how to balance performance with practicality and how to adapt our approach when faced with limitations.

It also gave us a better understanding of how real-world systems require thoughtful design decisions rather than just technically correct solutions.

What's next for Missing person Detection

In the future, we plan to extend the system to support real-time video processing using more optimized and scalable solutions. We also aim to improve the accuracy of the model by experimenting with more advanced face recognition techniques and larger datasets.

Our long-term goal is to make this system usable in real-world scenarios, where it can assist authorities and organizations in identifying missing individuals more quickly and efficiently.