Home Cases Machine Learning Eyewitness

Eyewitness

Software development with a focus on neural network systems and artificial intelligence

Technologies

Ruby on Rails, React

Industry

Machine Learning

Engagement Model

Dedicated Team

Platform

WEB

Team

3 members

Timeline

1 year

About the project

The platform we’ve built represents a unique online tool for every police officer conducting a real investigation process.

The range of project’s solutions includes the following services:

Challenges to overcome

Having discussed the project details with the client, we identified our primary task — building a visual and functional representation to make the client’s API appealing to the end-user.

At the stage of discussion of the future project, the client was equipped with the API used to identify the potential criminals. The development and management of inbound and outbound traffic of personal data required us to comply with the existing FBI CJIS regulations.

Next to this, stakeholders came up with another set of goals, including the analysis of the platform’s core functionalities and their mandatory integration into the web application. For example, our team has ensured the possibility for a police officer to quickly react to the system alerts about a potential suspect.

Initially, the project was intended for police departments, with a possible extension of functionality for security guards in the supermarkets as well as other security officials. In addition, the application’s backend was supposed to have direct access to several police databases, containing criminal lists with relevant information.

Our work

The platform that we’ve built in 3 months down the MVP plan is the unique tool for every police officer facing a live investigation process.

First, it builds up a specific case file consisting of the case ID, photos of potential suspects, additional information about the case, and physical address details of the committed crime.

Next, after you upload the required data and a photo, the results will return a list of registered criminals and the percentage score of the match between the photos of real criminals.

Finally, once all the steps of the search and identification have been completed, we can file an official report and construct a 6-pack of the wanted criminal.

Besides the crime investigation purposes, the control panel maintains the entire web platform. It can do the following:

Another important feature we’ve implemented is the possibility to upload a crime video for further investigation. Frame by frame, the program will extract the exact screenshot of a suspect and upload it so that a match can be done. The pictures’ quality was far from being ideal and could impede the identification of the suspect. Thus, our team decided to crop an average quality one to simplify the analysis process.

The high security level of the application is noteworthy. The application’s ecosystem was designed to comply with CJIS: it is hosted on the Microsoft Azure Government Could, all the stored data is encrypted in motion and at rest.

Business results

This application has been in the hands of 5 police departments and received rich positive feedback. The number of connected police databases is steadily increasing, with more potential criminals’ profiles being identified with high precision.

The clients are always satisfied with the produced results and have always been excited about the quality delivered by our team.

Currently, we are enthusiastically dedicated to the creation and development of the mobile version of Eyewitness! It’s still under construction, but we are sure it will be an excellent advancement to the existing platform that we have already presented and shared with you!