A survey of image labelling for computer vision applications

Abstract

Supervised machine learning methods for image analysis require large amounts of labelled training data to solve computer vision problems. The recent rise of deep learning algorithms for recognising image content has led to the emergence of many ad-hoc labelling tools. With this survey, we capture and systematise the commonalities as well as the distinctions between existing image labelling software. We perform a structured literature review to compile the underlying concepts and features of image labelling software such as annotation expressiveness and degree of automation. We structure the manual labelling task by its organisation of work, user interface design options, and user support techniques to derive a systematisation schema for this survey. Applying it to available software and the body of literature, enabled us to uncover several application archetypes and key domains such as image retrieval or instance identification in healthcare or television.

A Light in the Dark: Deep Learning Practices for Industrial Computer Vision

“…Just as humans use their eyes and brains to understand the world around them, CV attempts to produce the same effect so that computers can perceive and understand an image or a sequence of images and act accordingly in each situation. This understanding can be achieved by disentangling high-level, symbolic information from low-level image features using models built with the help of geometry, statistics, physics, and learning theory [7]. Driven by academic and industrial motives, grand advances have been made in several areas such as 3D model building, optical character recognition, motion capture, disease diagnostics and surveillance [8].…”

“…A field that significantly benefits from DL functionality is that of computer vision (CV). It seeks to automatically extract useful information from images to mimic human capabilities of visual perception [7,8]. On this basis, time-consuming and labor-intensive tasks like the recognition, detection, localization, tracking, and counting of objects can be supported more efficiently to save resources and relieve the burden of human workers [9].…”

Survey and Systematization of 3D Object Detection Models and Methods

“…Those are low-level features which are manually designed for the specific use case. On their basis, prediction models such as support vector machines (SVMs) can be trained to perform the object recognition task (Sager et al, 2021). However, the diversity of image objects and use cases in form of pose, illumination and background makes it difficult to manually create a robust feature descriptor that can describe all kinds of objects (Janiesch et al, 2021).…”

Obstacle Detection System for Agricultural Mobile Robot Application Using RGB-D Cameras

“…Images were manually labeled using the Computer Vision Annotation Tool (CVAT), which is a free, open-source, browser-based application. It provides convenient annotation instruments [46]. The second group is intended to test obstacle detection on maps.…”