Open Category Submission

312 – Forensics Smart System for Analyzing Crime Scene

Crime scene investigations often rely on manual processes and traditional forensic techniques, which can be time-consuming and prone to human error. This can lead to incomplete evidence documentation, inconsistent data analysis, and difficulty in examining complex crime scenes. Traditional methods may not adequately meet modern forensic training demands, such as accurate scene reconstruction, real-time evidence processing, and comprehensive data integration.

An AI-powered system for training crime scene investigators offers a promising solution by simulating investigative scenarios, virtually scanning crime scenes, and performing real-time analysis. With its ability to detect evidence and process data efficiently, the system ensures comprehensive and reliable training, reducing human error and enhancing the trainees' skills, preparing them to face real-world challenges in criminal investigations.

The idea for this project stems from our observation of the challenges faced by novice investigators during forensic training, particularly the significant gap between theoretical knowledge and practical application. Traditional educational methods rely on lectures and case studies without genuinely engaging trainees in investigative scenarios, which leads to numerous issues later when these investigators encounter real crime scenes. These limitations and challenges highlight the need for a new educational system that goes beyond conventional teaching methods and focuses on providing trainees with practical experience at minimal cost and risk.
This is where artificial intelligence plays a crucial role in bridging this educational gap by offering a solution that simulates crime scenes and allows investigators to interact with them. Our proposed project is not limited to virtual training alone; it also enhances analytical skills and decision-making while ensuring consistent training that reduces human errors. This approach integrates advanced technologies with forensic training, bringing about a significant transformation in investigator preparation, ultimately improving the efficiency of the educational process and yielding promising real-world outcomes.

--

The project structure is based on a web application built using Laravel, which serves as a platform for forensic evidence analysis, case management, and database interaction via MySQL. The system’s workflow can be divided into four phases. It begins with the login process for authorized users, followed by the data entry phase, where crime scene images and related information are uploaded to gather potential evidence.

Next, the Burhan system, utilizing object detection technology YOLOv8, classifies the uploaded images and identifies critical elements like fingerprints, weapons, blood stains, documents, and other evidence. Subsequently, the classification data is analyzed using a large language model (LLM). Finally, the AI model generates potential scenarios and hypotheses about the crime, Which in turn helps trainees to understand reality more broadly and think with more hypotheses.

Function Point → Technical Application (specific technical points) → Progress
Framework& Authentication → Laravel → Done
AI-Powered Object Detection → YOLOv8, OpenCV, PyTorch, TensorFlow (trained on Google Colab) → Done
AI-Assisted User Interaction → OpenAI GPT Model → Done
Image Processing & Preprocessing → OpenCV, Roboflow → Done
Data Storage & Management → MySQL → Done
UI Design → PHP, CSS, HTML, Canva, Bootstrap → Done

This project offers an innovative solution in the field of education and training for criminal investigations using artificial intelligence (AI) and Large Language Models (LLMs). The system integrates autonomous crime scene scanning, real-time object detection, and hypothesis generation regarding how the crime occurred. It also provides illustrative sketches and interactive videos to show how the crime unfolded based on the discovered evidence, helping trainees and law students visualize the sequence of events. The creativity lies in that the generative AI not only generates hypotheses but also creates visual representations to explain how the crime occurred. These sketches and videos help trainees understand the relationship between pieces of evidence and how they come together to form a comprehensive picture of the event. Additionally, LLMs contribute by generating detailed analyses and narratives about potential crime scenarios. By providing visual and interactive tools, the project enhances the effectiveness of training and assists trainees in understanding how to analyze evidence and generate hypotheses based on the available data. This way, the project offers a comprehensive learning experience that contributes to improving forensic investigation skills.

The "Burhan" project will ensure scalability and sustainability through the use of flexible cloud architecture, enabling automatic resource scaling to accommodate increasing data and user demand. By leveraging platforms like AWS or Azure, the system will maintain high performance as usage grows. The system will utilize innovative deep learning algorithms with continuous fine-tuning to ensure accurate crime scene analysis, while integrating generative AI for efficient data processing. Environmental sustainability will be prioritized through low-energy algorithms and resource-efficient design updates. Additionally, user-friendly interfaces will promote long-term engagement from investigators and researchers, with future integration capabilities to adapt to technological advancements. This approach guarantees that "Burhan" remains scalable, sustainable, and effective in the field of AI-driven crime scene investigation.

The solution we propose focuses on addressing a fundamental educational and social issue by enhancing the efficiency of criminal investigations and reducing potential human errors during training, ultimately promoting justice within communities. Moreover, it contributes to broader social goals, such as educational equality and equal learning opportunities, regardless of trainees' social and economic backgrounds. This is achieved by providing a unified virtual training environment for all trainees at minimal costs.

To measure the impact on society, we will use the following indicators:
1. The rate of error reduction in analysis and reasoning.
2. The level of trainees' satisfaction with the educational process through periodic surveys.
3. The improvement in the speed and accuracy of investigations and case resolution compared to the period before implementing the system.

Given the rapidly changing nature of our world today, we will adopt a scalable and continuously evolving development approach that remains flexible to accommodate new user requirements.

Personal Information Collection Statement (PICS):
1. The personal data collected in this form will be used for activity-organizing, record keeping and reporting only. The collected personal data will be purged within 6 years after the event.
2. Please note that it is obligatory to provide the personal data required.
3. Your personal data collected will be kept by the LTTC and will not be transferred to outside parties.
4. You have the right to request access to and correction of information held by us about you. If you wish to access or correct your personal data, please contact our staff at lttc@eduhk.hk.
5. The University’s Privacy Policy Statement can be access at https://www.eduhk.hk/en/privacy-policy.

• I have read and agree to the competition rules and privacy policy.