Core Concepts
Spannotation is an efficient tool for semantic segmentation in autonomous navigation, providing accurate masks quickly and effectively.
Abstract
Spannotation is an open-source tool developed for image annotation specifically in autonomous navigation tasks. It generates accurate segmentation masks for various environments like agricultural crop rows, off-road terrains, and urban roads. Unlike other tools, Spannotation achieves similar results in a significantly shorter time. The utility of the tool was validated through the training of a U-Net model with high accuracy and mean Intersection Over Union (mIOU). The simplicity, speed, and cost-free features have contributed to its adoption evident from its download count of 2098 since its launch. Future enhancements aim to broaden its application to complex navigation scenarios and incorporate additional automation functionalities.
Stats
Spannotation achieves similar results in about 6.03 seconds compared to other tools that require about 40 seconds for semantic segmentation annotation.
The U-Net model trained using Spannotation's generated masks achieved a validation accuracy of 98.27% and mean Intersection Over Union (mIOU) of 96.66%.
The download count of Spannotation stands at 2098 since its launch.
Quotes
"Unlike other popular annotation tools that requires about 40 seconds to annotate an image for semantic segmentation in a typical navigation task, Spannotation achieves similar result in about 6.03 seconds."
"The accessibility, simple annotation process and no-cost features have all contributed to the adoption of Spannotation evident from its download count of 2098 since its launch."
"Future enhancements of Spannotation aim to broaden its application to complex navigation scenarios and incorporate additional automation functionalities."