A solution to identify a brand-specific style

Type A set of scripts
TECHNOLOGIES NumPy, OpenCV, Python, TensorFlow
AREAS OF EXPERTISE Image processing, Computer Vision, Machine Learning, Neural Networks
TEAM 1 developer

The main goal of the project was to build a solution to identify a brand-specific style using images as a data source. Style is defined as a well-known color scheme, image composition and anything that may help one to affiliate an image with some brand. For example, with the BMW trademark. In other words, we were faced with the task of classifying images.

  1. Сars in the images, provided by the customer, might distract the neural network from its main focus (which is the background).
  2. The dataset was not big enough to achieve the required accuracy of 95% - it was only 2000 images when we needed at least 10 times more.
  1. We used the Darknet YOLOv3 algorithm to detect cars in the images; removed the rectangle bounding machine/machines – replaced it with a fully transparent color; cut a random rectangle if there were no cars in an image. It was necessary to provide the dataset homogeneity. Otherwise, images with and without rectangles would be split into 2 different classes and it will spoil the classification.
  2. We have performed data augmentation and created 10 new images from each input image with applying small transformations – rotating, scaling, horizontally flipping, filtering (changing brightness, contrast, colors of images). Now the dataset was rich enough to use it for training.

We have successfully built a model that steadily copes with its purpose. The accuracy value for our classification is 95% on the test data. In the future, with the appearance of new data (new BMW advertising campaigns, for example), it will also be possible to train a model on these images and improve neural network accuracy.

Business value

Such a solution could help young specialists in the Product design area with making sure their own projects are not plagiarizing any unique traits of well-known brands and are distinctive enough on its own.

It can also help young designers with finding out what well-known brands had the most influence when the new style was developed.

It can also help product design students to get acquainted more with some distinctive elements of branding for a better understanding of how to make their own projects more recognizable.

To put it simple, that solution could be used as ML for plagiarism search engines.

Concrete crack detection algorithm
Concrete crack detection algorithm
ML optimization of market conversion for a financial aggregator
ML optimization of market conversion for a financial aggregator