Criminal Identity Detector

It is my B.Tech final year project.

It is a Python Tkinter based desktop application that helps to identify criminal based on eye witness information. It draws the image of criminal based on user voice input and search the images from the database. It replaces the work of forensic artist and saves the time of searching the criminal. It does this quickly and accurately using deep learning algorithms.

It will be used by investigation teams because searching the criminal by uncleared image is very time consuming task. It need images of all the people of India which can be possible if this project is hand over to government. Peoples facial features change at different phases of life so it need image database creation for one time and will update these image using AI aging algorithm.

Application Architecture

Modules

Application Screen Shots

Home Page

Output - Text to Image by StackGAN

Output - Fetched image from database on comparison with generated image by Siamese Neural Network

Output - If image not found in database

Algorithms

Stack GAN

It is used to convert text as a input into image as a output.

Model Architecture

Siamese Neural Network

It is used to compare the two images to find similarity in them.

Model Architecture

Technologies

• Google Speech Recognition API
• PyAudio Library
• Auto PY to EXE