2019
Product design • Rightpoint Consulting
Role: Design strategy lead
Team: Product manager, Backend engineer (computer vision), 2 Frontend engineers (Android)
Platform: Android
Device: Samsung tablet
Background
The Home Depot (THD) approached Rightpoint to help test the efficacy of Computer Vision to validate digital planograms to support various merchandising tasks completed by store associates.
Currently, Home Depot creates plans for the layout of a bay. This includes product, POP (point of purchase) and fixtures. However, all three are stored in different sources and determining if a location is set correctly or if fixtures/POPs are missing is a difficult task.
Rightpoint and THD will work together to test Computer Vision solutions to help solve the POP and fixture problem. The current solution relies on paper planograms stored at each bay. This digital solution will render the current process obsolete, and ensure that merchandising associates can confidently conduct their roles in validating whether a bay is in compliance with the latest planogram documentation.
Project Goals
Leverage emerging technologies that enable Home Depot employees to quickly scan a bay or planogram for accuracy
Help determine if product, POPs (point of purchase) or Fixtures are in the right locations within THD Bays
Build a prototype application to help THD employees track products in bays
__________________________________
Project Retrospective
The computer vision app prototype can successfully identify POP signage with a high level of accuracy and draw boxes to show their location within a bay. Computer vision has been proven to be functional within the Home Depot retail stores during onsite testing. Computer vision code has been written in a platform portable methodology. MET supervisors understand the need for such a tool and how it works through user testing at Home Depot stores.
A set of onboarding screens were introduced for first time users. The steps required a user to identify a bay within a department by either manually inputting a bay number or scanning a bay tag. Once the user had identified the bay, they would use the tablet’s camera to take a stitched panoramic photo of the bay. Once the photo was successfully taken, the application would use computer vision to scan the bay against the digital planogram on file and report back any discrepancies.
Summary of Onboarding Steps
Scan the Bay ID tag on the upright beam. Match the bay image and POG on screen with the bay in front of you. Take a panorama of the bay. Follow on-screen prompts to perform this photo capture. Confirm correct, missing, and out-of-date signage on screen using the picture. Review the results, submit a report, and order replacement signage.
On this dashboard page, we can see a heads up view of store rank and overall compliance, as well as a ranking of compliance levels of bays in your store. The user has the ability to drill down into a department view for your store and the average bay compliance level. The user can drill in even further and see bay compliance for each department in your store. Additionally, the supervisor can review daily tasks, and the results of finished assignments (reports and orders).