Spannotation: Enhancing Semantic Segmentation for Autonomous Navigation with Efficient Image Annotation

TL;DR

Spannotation tackles the labeling bottleneck in semantic segmentation for vision-based autonomous navigation by providing a fast, free triangle-based annotation tool that produces binary drivable-area masks. The method relies on three user clicks to define a triangular region, which is rasterized into a binary mask using standard geometric primitives and scanline fill, with a Python-friendly implementation built on OpenCV, NumPy, and related tools. Training a U-Net on Spannotation-generated masks yields high performance (val. accuracy 98.27% and mIOU 96.66%), across agricultural, off-road, and on-road scenarios, validating the masks as effective training data. The tool shows strong adoption (over 2k downloads) and competitive advantages in setup speed and navigation-focused applicability, with future work aimed at handling more complex scenes, adding automation, and improving usability and integration.

Abstract

Spannotation is an open source user-friendly tool developed for image annotation for semantic segmentation specifically in autonomous navigation tasks. This study provides an evaluation of Spannotation, demonstrating its effectiveness in generating accurate segmentation masks for various environments like agricultural crop rows, off-road terrains and urban roads. 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 tools utility was validated through the utilization of its generated masks to train a U-Net model which achieved a validation accuracy of 98.27% and mean Intersection Over Union (mIOU) of 96.66%. The accessibility, simple annotation process and no-cost features have all contributed to the adoption of Spannotation evident from its download count of 2098 (as of February 25, 2024) since its launch. Future enhancements of Spannotation aim to broaden its application to complex navigation scenarios and incorporate additional automation functionalities. Given its increasing popularity and promising potential, Spannotation stands as a valuable resource in autonomous navigation and semantic segmentation. For detailed information and access to Spannotation, readers are encouraged to visit the project's GitHub repository at https://github.com/sof-danny/spannotation

Figures (6)

• Figure 1: General overview of the Spannotation process. Spannotation can be used for either single image (option 1) or multiple images in a folder (option 2) where N is the total number of images in the folder. There is no order to which the three (3) points can be selected on the image.
• Figure 2: Schematic representation of the U-Net architecture.
• Figure 3: Original images and corresponding masks generated by Spannotation for different scenarios: agricultural crop rows, off-road, and on-road.
• Figure 4: U-Net training results over 33 epochs: (a) Validation loss, (b) Validation accuracy, and (c) Mean Intersection over Union (mIOU).
• Figure 5: Comparison of segmentation results: (a) Original images (b) spannotation masks, and (c) U-Net generated segmentation.