Counterfeit Product Detection Using Smartphone

Abto Software keeps pace with the latest technology developments and trends. For two years our company has been working with a well-established European partner on the algorithms in the image processing technologies.

Our App detects fake consumer goods with 95% accuracy

Team and Technologies

Our software engineers conducted a scientific research and developed a library to solve the issue of counterfeit detection. Fraud is an increasing problem, and also a concerning question for manufacturers and retailers all over the world.

They built a commercially available mobile application for Android and iOS which performs authenticity analysis of data comprising image data and associated meta-data.

Technologies employed

MATLAB, Python, Qt, OpenCV, Android, IOS, image processing, information theory

Functionality of Counterfeit Product Detection App

The application recognizes content from images and transfers tags to content publishers and providers to improve the consumer experience. It’s a value-added tool which performs image quality assessment to distinguish between legitimate and falsification patterns in a real time mode. Despite all the development complexity, the application is designed to be particularly user-friendly. The consumer only ensures the actual presence of a real image and the engine will start doing the matching with no pre-requisitions.

The app identifies multiple objects by leveraging algorithms with hundreds of visual aspects. The tool produces an estimate of the image authenticity by selecting the region of interest and overlaying an image on the object in order to analyze visual content and identify matching images between a query image and a reference database.

This powerful image matcher uses a smartphone’s high resolution camera to fulfill real-time image recognition for Android and iOS and gives matched results with performance speed of less than a second.

Key features of the Counterfeit Detection app comprise:

• Real-time image processing
• Invariance against rotation, illumination, and flip
• Automatic calibration
• Accurate matching with almost any image
• Easy to setup, manage the image database
• Batch mode
• Possibility to load images from the mobile resources or from an URL
• Results  ready in less than a second
• Highly intuitive and configurable

The application comprises input means arranged to receive data from numerous sources and processes the imaging data for the purpose of determination whether it meets a fraud criterion, and if it does, to generate a fraud indicator output.

Challenges

The most challenging part of the project was to achieve the highest precision level which simultaneously reduces the algorithm’s execution speed. With 95% of accuracy level our meticulous investigation turned into a proof of concept.

There was one more challenge which our team turned into a noticeable benefit. They obtained the highest quality of search outcome by extracting key visual features, excluding  shape and color from an image and used this information, along with contextual data, similarities analysis  for more meaningful results.

Benefits of the application

There are numerous benefits for both human and commercial potential in mainstream life.

• Robustness against blur, crop, perspective
• Reduction of  the number of mistakes when taking a photo
• Computationally affordable algorithm
• Large-scale database
• Highest level of usage simplicity while providing the right functionality
• Enhances security level, delivers operational efficiencies and protects customer’s brand

This highly competitive application reveals valuable information that can be efficiently used to segregate bona fide from fake. The app runs on the modern top mobile devices and can be applied to the spheres requiring search and identification of the matching images.

Having a resource to clarify authenticity  means a vital difference between a wise investment in quality and excessive spending.