Strawberry Detection using Mixed Training on Simulated and Real Data

TL;DR

This paper explores the use of combined simulated and real images to improve strawberry detection accuracy, demonstrating that data augmentation with simulated images can enhance model performance in agriculture.

Contribution

It introduces a mixed training approach using simulated and real data for strawberry detection, showing improved accuracy over using real data alone.

Findings

Mixed training with simulated data slightly improves detection accuracy.

Simulated data can effectively augment real datasets in agricultural object detection.

The approach reduces the need for extensive real labeled data.

Abstract

This paper demonstrates how simulated images can be useful for object detection tasks in the agricultural sector, where labeled data can be scarce and costly to collect. We consider training on mixed datasets with real and simulated data for strawberry detection in real images. Our results show that using the real dataset augmented by the simulated dataset resulted in slightly higher accuracy.