Planogram

E-commerce

Overview

Planogram compliance for detecting missing or misplaced items. A retail store frequently has to cope up with demand & much larger stores need to employ staff just to check up on inventory, products placed on the shelves & get items in place misplaced by customers, a very frequent occurrence & annoyance for storekeepers. With that taken into account, we have a solution to get instant updates on if a shelf is empty, the item is missing & our end-to-end system gets real-time updates from inventory linked to the database & assigns someone instantly to fix the situation.

Technology

Computer Vision
Deep Learning
Machine Learning

Language

Python

Key Technical Challenges:

Having varying products on the shelves, we had to make sure we make generalized model
Data provided to us did not cover all the use cases & we had to get new data & annotate them in order.
Establishing a streamline pipeline with a Deep learning model was a bit difficult.

Business + Technical Points:

Needed real-time updates & notifications from the store
Role-based access to the system & Modification of positioning of shelves should not affect the detection
Surveillance quality was good but modernization may happen in future, the system should cope with it.
Resource consumption in terms of model should be optimal & should be scalable when need arises for expansion.