Artificial intelligence powered crime scene analysis service
Crime remains a major concern in modern culture. As a result, emphasizing prevention and ensuring prompt justice delivery are critical. In criminal investigations and law enforcement, forensic science plays a critical role. However, typical approaches in this field frequently rely on physical proced...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-12-01
|
| Series: | MethodsX |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2215016125002766 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Crime remains a major concern in modern culture. As a result, emphasizing prevention and ensuring prompt justice delivery are critical. In criminal investigations and law enforcement, forensic science plays a critical role. However, typical approaches in this field frequently rely on physical procedures, which are inefficient and likely to increase human mistakes, hampering the timely administration of justice. To tackle these issues, artificial intelligence skills were integrated into investigation processes. The suggested novel web application aims to help forensic investigators streamline crime scene investigations by automating important portions of the procedure.The system has three main functions: • fingerprint reconstruction, • weapon detection, and • human activity recognition.Its easy-to-use interface allows officials to upload images or videos from the crime scene in order to assist in reconstructions and detections. The fingerprint reconstruction model built using autoencoders outputs better partially covered fingerprint images with a validation loss of 0.0477 and a step loss of 0.0487, which helps detect the persons responsible for the event. Furthermore, the model for object detection, YOLO NAS, plays an important role in recognizing weapons that might be present at the scene, with an mAP of 77.8 %, while human activity detection techniques such as VGG16 have a total accuracy of 98.21 %. |
|---|---|
| ISSN: | 2215-0161 |