Consulting Training Blog Careers About Us Contact Us

Case Study

  • Client: Computer Vision-based Catalog Matching for a Multi-Billion Dollar Electronics Component Distributor
  • Project: Identify an electronics component based on an image to lookup replacement parts
  • Country/Region: Minnesota-based, International distribution

An international electronics distributor recently engaged Intertech for technical support on a special project involving neural network-powered computer vision. The key idea behind the application is to allow customers to take a picture of an electronics component and receive a matching component from the distributor’s catalog of over 8 million products. Currently, human customer support specialists field queries to find matching parts using their expertise. Neural networks could automate much of this process, making it faster, cheaper, and more accurate to find matches for customer inquiries.

Technologies Used

  • Python
  • Amazon Web Services
  • Docker
  • PyTorch
  • Tensor Flow
  • imgaug

Challenge

Every day, the client’s 8000 employees fill three jumbo jets with electronics components for customers around the world. Speed and accuracy are high priorities for the client’s business. Indeed, many of their customers need electronics components shipped overnight to meet production quotas or repair downed systems. In turn, those quickly delivered components need to fit perfectly and match the specifications of the given design.

Currently, the client employees a team of parts specialists who are experts in electronics components. These specialists use their years of experience to identify parts and suggest replacements when customers have inquiries. However, this process can still be time consuming, especially for rare parts or components with highly specific applications. In those cases, parts specialists must comb through hundreds or thousands of entries in the client’s catalog of over 8 million components.

In order to speed up the parts discovery process, the client has commissioned a special team to investigate the potential of using neural networks to classify images of parts. Ultimately, they hope that the customer can take a picture of the component they need to replace and the neural network will be able to match that component to likely replacement components in the catalog.

Process

The client engaged Intertech to source an expert in neural networks and data infrastructure to alongside their in-house expert on the small exploratory team. Intertech’s expert consultant is one of a team of three technologists working on the project.

Broadly, the process for developing a new solution falls into two key tasks. First, the team must create, train, and test a neural network for image recognition of electronics components to a high degree of accuracy. Second, the team must lay the infrastructure for storing, retrieving, and processing hundreds of millions of training images for the neural network to learn from.

In order to successfully identify an electronic component from a photo, the team began by taking images of a small set of test components from all angles. For each component, the team has 96-192 images of that component rotated in all possible directions. From there, they use the imgaug Python library to simulate various levels of distortion, saturation, brightness, and blur in the photos. Ultimately, the team uses PyTorch and TensorFlow models written in Python to create a neural network that then consumes the training images in order to learn. Tweaking and honing those models over time to achieve maximum accuracy is a major challenge of this project.

Additionally, the team must find a way to store hundreds of millions of training images for all the components in the client’s catalog. Originally, they used Amazon Web Services for storage and processing, but they found it was too expensive to be feasible. So, the team ordered in custom hardware with high-end GPUs to do the image processing in-house.

Finally, the team is also looking for ways to optimize their training speed. It could take the neural network months to process a million images. A few thousand images already take several hours. As such, every small bit of efficiency the team can gain will matter to how quickly the model can be trained and retrained if new data or a better model becomes available.

Learn More About Intertech Consulting

Result

This project is ongoing, but early tests on a small subset of components have shown promise. Of course, scaling the solution to an 8 million part catalog still remains a challenge that the team is working on. Additionally, reducing processing times for images will be a key part of scaling the solution.

Due to the success of early trials, the tool could be something that rolls out in phases. The current parts specialist team could use the neural network to help identify some categories of components, with capabilities increasing over time.

Ultimately, this project represents a major step forward and a transformative shift in the way the client will provide customer service. Eventually, customers may have their inquiries answered instantly by just submitting a photo from their smartphone. This frees parts specialists to spend less time combing the catalog and more time making substantive recommendations to customers about the best components for their needs.

Our Story: Intertech

From the day it was founded in 1991 by local entrepreneur Tom Salonek, Intertech has been a company with an important difference: unwavering commitment to customers, employees and the broader community through excellent work, smart workplace and financial management, and creative philanthropic involvement.

Learn More About Intertech Consulting