Flow Diagram of the operartion day

Personas Workday

For a perfect workday operation, we rely on two professionals: the

data supervisor and the field team. Their roles are as follows:

Data Supervisor: This person is responsible for performing and reviewing

registrations, approving images, and conducting and monitoring training. It

acts in the back office, and may delegate part of its functions, as appropriate.

Field team (merchants): Using the mobile application, these persons go to

the sales locations to capture data to execute the planograms.

General flow discription

For automatic recognition, it is necessary to register products in the IRE

through the backoffice. Within these products, image catalogs are

registered, representing different forms of seasonal variations and specific

packaging.

It is important to note that the inclusion of new products or packaging

requires retraining of the recognition model. If these new products are not

inserted into the planogram, they will not be part of the recognition analysis

performed by the system.

The automatic recognition is based on machine learning. The images in the

catalogs are used as examples for the system to learn to recognize the

products.

After making changes to the products and catalogs, training is required to

improve recognition, ensuring that the model is up to date with the new

additions.

When the data supervisor or field operation person synchronizes on the

mobile application, the planograms and product recognition are updated.

When capturing an image, the system will try to automatically recognize the

product, considering only the products present in the planogram.

It is important to ensure that all new products are correctly registered and

included in the planogram so that the system can perform the correct

recognition analysis. Incorrect or missing identifications can be edited to

improve the accuracy of the system.

Edited images go through approval in the backoffice to be added as new

examples in the catalogs. The system improves its recognition as new

images are approved and retrained, following good practices. The quality of

automatic detections can be monitored through dashboards.