Use Case and High-Level Description

A product detector based on the SSD-lite architecture with MobileNetV2 as a backbone for self-checkout points of sale-related scenes. The network can detect 12 classes of objects (sprite, kool-aid, extra, ocelo, finish, mtn_dew, best_foods, gatorade, heinz, ruffles, pringles, del_monte). Labels 0 and 1 are related to background_label and undefined correspondingly.

Example

Specification

Performance

Inputs

1. Name: input , shape: [1x3x512x512]. An input image in the format [BxCxHxW], where:

   • B - batch size
   • C - number of channels
   • H - image height
   • W - image width

   Expected color order: BGR.

Outputs

1. The net outputs a blob with the [1, 1, N, 7] shape, where N is the number of detected bounding boxes. For each detection, the description has the [image_id, label, conf, x_min, y_min, x_max, y_max] format, where:
   • image_id - ID of the image in the batch
   • label - predicted class ID
   • conf - confidence for the predicted class
   • (x_min, y_min) - coordinates of the top left bounding box corner
   • (x_max, y_max) - coordinates of the bottom right bounding box corner

Legal Information

[*] Other names and brands may be claimed as the property of others.