InvestiVision

Inspiration

In forensic science, every detail counts. From crime scene images to pieces of evidence, hidden details can make or break a case. However, traditional image processing methods often lack the precision and depth required for forensic investigation. InvestiVision was inspired by the need to enhance image clarity, reveal critical evidence, and identify objects in crime scenes with the help of AI—all to support forensic professionals in their work.

What it does

InvestiVision enhances crime scene images using advanced image processing techniques. With features like brightness and contrast adjustment, noise reduction, sharpening, grayscale conversion, and object detection powered by YOLOv5, it enables forensic analysts to reveal subtle details and identify key objects in scenes. The app’s interface is user-friendly and designed to improve the clarity of crime scene images, uncover hidden information, and streamline forensic investigation.

How we built it

The project was built using:

Python and OpenCV for image processing tasks, including noise reduction, edge detection, and contrast enhancement. YOLOv5 for object detection, specifically for identifying items that may be relevant in a crime scene. Streamlit as a frontend to create an interactive and user-friendly web app. PIL (Python Imaging Library) for image conversions, downloads, and rendering. Custom Styling (CSS) to give the app a clean, professional look.

Challenges we ran into

High Processing Demand: Managing the computational load of image processing and object detection tasks, particularly for high-resolution images. Technical capabilities - deblur-gan was highly complex. Integrating YOLOv5 with Streamlit: Ensuring that YOLOv5’s model worked efficiently with Streamlit's live user interface. Real-time Adjustments: Providing smooth real-time updates when adjusting image parameters without lagging. Balancing Complexity and Usability: Designing an interface that remains easy to use while providing advanced features.

Accomplishments that we're proud of

Developing a Professional and User-Friendly Interface: InvestiVision’s clean, streamlined design enables easy navigation and usability for professionals. Successful Object Detection Integration: Implementing YOLOv5 allowed us to automate the detection of key objects in crime scenes, enhancing forensic analysis. High-Quality Image Enhancements: Achieving high-quality enhancements like noise reduction and sharpening, preserving details critical for analysis. Making friends and valuable connections. Networking amongst peers

What we learned

The Importance of Optimization: Efficient processing is essential for real-time applications, especially with computationally intensive tasks. Seamless Integration of AI Models: Combining machine learning models with user-friendly interfaces requires careful planning and optimization. User-Centric Design: Designing with end-users in mind, particularly forensic professionals, is crucial to ensure the app’s usability and effectiveness.