Image Recognition is a tool that uses artificial intelligence to compare implementation photos with a planogram. This feature allows automatic detection of differences between the planogram and the actual shelf display and evaluates the quality of implementation as a percentage, complemented by a visual indicator.

Recommended Photo Quality

For optimal Image Recognition results, we recommend using photos with a resolution of approximately 10 MPx.

The quality of the uploaded photo can be found in the photo table (in the Recent Photos / Implementation Photos section) in the Resolution column.

Table of uploaded photos with the indicated resolution

Where is the feature Image Recognition available?

• • Quant Web Application 
  • Quant Desktop Application

Image Recognition in the Web Application

Image Recognition is an add-on feature to the implementation photos linked to planograms. To activate the feature, click the Show in Planogram icon for a specific photo in the Photos section. Once activated, a comparison window of the implementation photo with the corresponding planogram will open.

Show in Planogram Icon

Quant automatically compares individual products in the photo with those in the planogram and evaluates the quality of the implementation.

Result indicators:

• • Green box marks products correctly placed according to the planogram.
  • Red box marks products that should not be on the shelf.

In the top right corner of the screen, a visual indicator (in the form of a colored smiley face) shows the overall quality of the implementation. Hovering over the indicator displays the exact Implementation Quality Rate Percentage.

Implementation Quality Assessment Range is shown using a color-coded indicator based on default values:

Image Recognition and Implementation Quality Percentage

Detailed Analysis of Implementation Quality

For a detailed evaluation of recognition results, Quant offers the following tools available in the top-right corner as icons:

• • Annotations Icon - shows or hides additional notes and information for individual products on the shelf detected during recognition.
  • Bounding Boxes Icon - toggles the highlighting of frames on the picture. Disabling it can help you better view the original image without visual overlays.
  • Edit Crop Box Icon - allows you to adjust alignment and crop the picture for more accurate comparison with the planogram, for example when the picture was taken at an angle or the shelf is not properly centered.

Icons for Detailed Analysis of Implementation Quality

The Image Recognition function is also available for individual stores, allowing them to verify the display status immediately after uploading a photo for implementation. This feature can serve both as a feedback tool and as a motivator for better and higher-quality planogram implementation directly at the stores.

To access detection results and to be able to edit crop box, the corresponding user web rights must be activated (Show Detection Results and Edit Detection Crop Box).

Web rights - Show Detection Results and Edit Detection Crop Box

Image Recognition in the Desktop Application

In the Quant desktop application you can find uploaded photos in the Recent Photos / Implementation Photos section. After opening this section, you can sort the photos according to various attributes set in the table - in our case, the photos are sorted by Implementation Quality. Clicking on individual rows displays a detailed preview of the photo and the corresponding planogram.

Recent Implementation Photos

How can you manually crop a photo in Quant?

If the photo was taken at an angle or the shelf is not properly detected, you can adjust it manually as follows:

1. Activate the Select Polygon icon to edit the photo's crop area. Quant will offer to adjust the polygon corresponding to the detected shelf area.

Icon Select Polygon

1. Move the individual corners of the polygon as needed and click to confirm the adjustment of each corner.

Editing and Cropping the Photo

1. After adjusting the crop, activate the Product Detection button to re-run automatic recognition and implementation evaluation.

Re-running Automatic Recognition and Implementation Evaluation

When is product detection visible in photos?

If the Image Recognition module is activated afterward, and implementation photos already exist in the database, these photos will not be automatically detected immediately.

For retroactive detection, you can use the Product Detection button in the desktop app, including the option for bulk detection of multiple photos at once. Similarly, implementation photos viewed in the planogram detail on the web will also respond by triggering detection.

Newly added implementation photos will be detected automatically from then on.

Bulk Detection

If the user has the Task management module activated, it is possible to create a New Task directly from the photo preview to notify the store about an incorrect or insufficient implementation.

Depending on user rights, a button Approve is available to certain users, allowing them to approve the implementation of the selected planogram.

The Approve and New Task buttons are not visible to everyone - their availability depends on the user's rights settings and activated modules.

Approve and New Task buttons

More information about uploading photos to planograms can be found in the manual Implementation photos for planograms.