Towards Context-Driven Approaches to Visual Affordance Recognition : A Comprehensive Survey

With the rise of robotics, affordance detection has become a key element for improving and strengthening human and robot object interactions. Thanks to recent progress in deep learning, affordance detection has become more precise and versatile. However, with the increasing use of Large Language Models (LLMs) and Visual Transformers (ViTs), the research focus shifted from clearly defined classes and dynamics to wider applications. Due to these new applications, context-based affordance recognition has gained popularity. However, to the best of our knowledge, there is no unified work around the comprehension and description of these advances in a context-based affordance detection. Therefore, this paper aims to review recent contributions in the field of contextual affordance detection, to provide solid foundations to the community, highlighting current works and future trends. This paper first introduces the context types and different tasks for affordance recognition. Then, various context-based methods for affordance recognition are presented, classified according to the context they are based on and their learning paradigms. Next, relevant datasets are presented with the respective methods’ performances on these datasets. Finally, this paper provides a discussion around the presented results and the trends observed through this survey, followed by a summary of the challenges and open questions to be faced in the future.