Verifying compliance of sold, missing, misplaced items in real shelves to the ideal shelves is a costly task. This poc is related to planogram compliance for detecting missing or misplaced items.
- Planogram refers to the arrangement of products in store shelves
- Our model compares the two images of the shelf that is the one when the shelf was arranged and the one after which some items were sold or misplaced
- The model uses computer vision architecture to analyse images at the pixel level
- Every pixel has its own rgb value. So model compares rgb values of a pixel in master image with the rgb value of that in the test image.
- Using certain algorithms, our model successfully detected the missing or misplaced items from the shelf.
Key values / Achievements
- We were able to distinguish between the master image with the test image with 90% accuracy.
- Bounding box is created for the missing objects.
- The output image is as shown below: