Searching for Efficient Multi-Scale Architectures for Dense Image Prediction
This work addresses the challenge of generalizing meta-learning methods to dense image prediction tasks like scene parsing and segmentation, offering improved efficiency and performance for computer vision applications.
The paper tackled the problem of designing efficient neural network architectures for dense image prediction tasks, achieving state-of-the-art performance with 82.7% on Cityscapes, 71.3% on PASCAL-Person-Part, and 87.9% on PASCAL VOC 2012, while reducing parameters and computational cost by half compared to previous systems.
The design of neural network architectures is an important component for achieving state-of-the-art performance with machine learning systems across a broad array of tasks. Much work has endeavored to design and build architectures automatically through clever construction of a search space paired with simple learning algorithms. Recent progress has demonstrated that such meta-learning methods may exceed scalable human-invented architectures on image classification tasks. An open question is the degree to which such methods may generalize to new domains. In this work we explore the construction of meta-learning techniques for dense image prediction focused on the tasks of scene parsing, person-part segmentation, and semantic image segmentation. Constructing viable search spaces in this domain is challenging because of the multi-scale representation of visual information and the necessity to operate on high resolution imagery. Based on a survey of techniques in dense image prediction, we construct a recursive search space and demonstrate that even with efficient random search, we can identify architectures that outperform human-invented architectures and achieve state-of-the-art performance on three dense prediction tasks including 82.7\% on Cityscapes (street scene parsing), 71.3\% on PASCAL-Person-Part (person-part segmentation), and 87.9\% on PASCAL VOC 2012 (semantic image segmentation). Additionally, the resulting architecture is more computationally efficient, requiring half the parameters and half the computational cost as previous state of the art systems.