Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecture
This work addresses multiple vision tasks with a unified approach, offering a versatile solution for researchers and practitioners in computer vision.
The paper tackled depth prediction, surface normal estimation, and semantic labeling in computer vision using a common multi-scale convolutional architecture, achieving state-of-the-art performance on benchmarks for all three tasks.
In this paper we address three different computer vision tasks using a single basic architecture: depth prediction, surface normal estimation, and semantic labeling. We use a multiscale convolutional network that is able to adapt easily to each task using only small modifications, regressing from the input image to the output map directly. Our method progressively refines predictions using a sequence of scales, and captures many image details without any superpixels or low-level segmentation. We achieve state-of-the-art performance on benchmarks for all three tasks.