Recent Advances of Continual Learning in Computer Vision: An Overview

TL;DR

This survey addresses continual learning for computer vision by formalizing problem definitions, evaluation metrics, and a comprehensive taxonomy of methods. It covers regularization, distillation, memory-based, generative replay, and parameter isolation strategies, along with their combinations and applications across CV tasks. Performance comparisons across online learning, few-shot incremental learning, semantic segmentation, object detection, and re-identification reveal task-dependent strengths and current state-of-the-art trends. The discussion outlines future directions including cross-domain applications, unsupervised and multi-problem settings, and efficiency considerations for real-world deployment.

Abstract

In contrast to batch learning where all training data is available at once, continual learning represents a family of methods that accumulate knowledge and learn continuously with data available in sequential order. Similar to the human learning process with the ability of learning, fusing, and accumulating new knowledge coming at different time steps, continual learning is considered to have high practical significance. Hence, continual learning has been studied in various artificial intelligence tasks. In this paper, we present a comprehensive review of the recent progress of continual learning in computer vision. In particular, the works are grouped by their representative techniques, including regularization, knowledge distillation, memory, generative replay, parameter isolation, and a combination of the above techniques. For each category of these techniques, both its characteristics and applications in computer vision are presented. At the end of this overview, several subareas, where continuous knowledge accumulation is potentially helpful while continual learning has not been well studied, are discussed.

Figures (5)

• Figure 1: General trend of the number of papers on continual learning in computer vision published in top-ranked conferences during the past eight years. The plot shows consistent growth in recent literature.
• Figure 2: A taxonomy of continual learning methods.
• Figure 3: Illustration of EWC kirkpatrick2017overcoming, which bounds the model parameters to update to the common low loss area among tasks, instead of the low loss area of only the new task.
• Figure 4: Illustration of LwF li2017learning, which stores a copy of the previous model's parameters before learning the new task, and uses the response of that copied model on the data instances from the new task as the target for the previous tasks' classifiers during the learning of the new task, while using the accessible ground truth as the target for the new task classifier.
• Figure 5: Illustration of PathNet fernando2017pathnet, which freezes the model parameters along all paths selected by the previous tasks and reinitializes and retrains the remaining model parameters following the same process to select the best path.