Back to Results...

This research is performed within the scope of the project "Object recognition in 3D scenes based on depth and image data" funded by Fast Global Solutions, Inc., Glenwood, MN, USA.
This page presents results of a CNN-based algorithm for package detection and localization in depth images in different scenes, and scenes with consecutive removal of packages. Below you can see photographs of sample scenes, and segmentation results in the corresponding depth image below each photograph. The pages marked with the same color represent a single box.
Read more details in A Hybrid CNN-DSP Algorithm for Package Detection in Distance Maps

The labeled distance map dataset created within the scope of this project is publicly available for research purposes, and can be downloaded from here.

Cite as: E. Vasileva and Z. A. Ivanovski, "A Hybrid CNN-DSP Algorithm for Package Detection in Distance Maps," in IEEE Access, vol. 11, pp. 113199-113216, 2023, doi: 10.1109/ACCESS.2023.3324058.