DeepScanner: a Robotic System for Automated 2D Object Dataset Collection with Annotations

TL;DR

DeepScanner introduces an automated robotic system that significantly accelerates and improves the accuracy of 2D object dataset collection and annotation, reducing manual effort and enhancing neural network training quality.

Contribution

The paper presents a novel robotic system for automatic dataset collection and annotation, achieving 240-fold faster labeling and 13-fold higher accuracy than manual methods.

Findings

Data labeling speed increased 240 times

Annotation accuracy improved 13 times

Neural network trained on automated dataset performs comparably

Abstract

In the proposed study, we describe the possibility of automated dataset collection using an articulated robot. The proposed technology reduces the number of pixel errors on a polygonal dataset and the time spent on manual labeling of 2D objects. The paper describes a novel automatic dataset collection and annotation system, and compares the results of automated and manual dataset labeling. Our approach increases the speed of data labeling 240-fold, and improves the accuracy compared to manual labeling 13-fold. We also present a comparison of metrics for training a neural network on a manually annotated and an automatically collected dataset.