INKuisitor

Image processing and ML based, forensic offline signature verification Web Application

Abstract :

Biometrics (or biometric authentication) refers to the method of identification of humans by their characteristics or traits. Biometrics is employed in computing as a sort of identification and access control which is one among the foremost secure methods to ensure human’s privacy. The application uses a Convolution Neural Network (CNN) for distinguishing between forged and genuine signatures through a Siamese Neural Network. With an input of three signature images to create a new signature profile , store in database for later verification requests and a single input signature image used for verification of person’s identity.

Available Scripts

In the project directory, you can run:

npm install

Install dependencies (Important Node Modules to start the project)

npm start

Runs the app in the development mode.
Open http://localhost:3000 to view it in the browser.

Links :

INKuisitor's DEMO

INKuisitor's Documentation

INKuisitor's Presentation

INKuisitor's Backend

INKuisitor's Machine Learning Model

To-Do List:

• implement UI and Layout
• Divide application into pages
• Enable routing (connect react router).
• integrate with Backend server

• Verify Page:

• divide verify page into components
• implement verify page form component
• implement dynamic-content dropdown list(using React Select)
• implement signature pad to allow user to sign on the go (using mouse cursor or touch pen)
• implement uploading a file button (allow user to upload an already existing signature image on hard disk)
• implement option selection (using React Select) to allow user to chose between signing either on signature pad or uploading a signature image
• implement dynamic reactive component , changes upon getting response from server from form component to gauge component
• implement gauge to show the percentage of the similarity between the provided signature and the stored on database one , and showing wether the signature is forged or geniune based on a pre-defined threshold (React-gauge)
• handle side effects , fetch the profiles array from the server
• handle sending POST request to server from verify component , and taking response put inside the gauge (percentage of similarity)
• Verify Page Block Diagram :

• Upload Page:

• divide upload page into compoennt
• implement upload page's form component , allowing user to interact with a form to add a new client signature profile, same as the previous form.
• implement dynamic reactive component , changes upon getting response from the the server that account is successfully created (react-success-indicator) , changing on the UI from form componet to the success indicator message
• handle sending a mulitpart POST request to the server ,contains name of the client profile and attached a FileList object , taking response to indicate successful creation of a new profile
• Upload Page Block Diagram :

• Others:

• implement Home page ,divide it into components , write a short brief about the application
• implement other pages in the application , about and legal
• Enhance Styling of the Application