Technology

Data Science

Industry

Retail

Brand Performance Analysis using Product Recognition

A blend of Deep Learning & Computer Vision is re-defining the Brand Performance Analysis in Retail Environment.

Client Profile

Our client is one of the leading chains of retail outlets all over the world. With over a hundred stores and a million plus products, they are embracing technology at a faster pace to give the best of experience to the external and internal customers to the brand producers.

Overview

Shopping in person, wandering through the infinite racks of a supermarket might give a relaxing time to the customers. But with technology evolution, an automatic inventory system to report the availability of products along with the demand prediction and refilling of the shelves, is a bliss for the Brand. The demand prediction will be useful for analysis of buying pattern of the customers as against the host of products from the completion in the similar space

Key Challenges

Product recognition in the retail stores poses many challenges such as identifying different products under various categories. Availability of the Intra & Inter class variability and variety along with the specific catalogues on promotional schemes and offers are dynamic in nature. Elimination of misplaced products, empty spaces on the visual merchandise and categorisation of products on display are the other challenges any Brand will face in a retail environment.

Our Approach

Capturing the products on display, pattern matching with the catalogue, detecting the empty spaces, frequency of the sales for each product under all categories and analysing customers buying pattern has been the key metrics for building this solution.

Our Solution

Using image detection techniques with a blend of deep learning has solved the most critical problem faced by the Brand. Customized neural networks have been built by extracting the features of the captured images and training the machine for real time dashboards. Real-time analysis of the buying pattern of the customers will provide with insights on where the product stands against the competition that will result in informed choices on the pricing pattern and positioning of products in the retail space for the brand.

Technology Used

  • Deep Learning
  • Python 3.6
  • Open CV
  • AWS
  • Django
  • Label IMG toolbox
  • Custom Neural Networks

 

  • Darknet
  • TensorFlow
  • Keras
  • Chainer
  • Mxnet
  • ONNX

Results

  • Easily recognises Specific Brand demands in each localized retail outlets.
  • Sales Analysis results in tremendous growth of a product.
  • Helps the brand to innovate the promotional strategies.
  • Elastic dynamic product processing for detection.
  • Less computational complexity as custom neural networks are implemented.
  • Any volume of products can be trained and detected.

Download Brand Performace Analysis Case Study

FREE DOWNLOAD

Send download link to:

Other Case Studies