Multi-scale depth classification network for monocular depth estimation

In addition to the RGB information of an image, depth information is the most critical. Monocular depth estimation is an effective method for predicting depth from RGB images. First, we propose a multiscale classification network that transforms the predicted depth values into a nonlinear combination problem between multiple depth interval values. Based on the correlation of the depth information, a depth classification network was used to determine the results of the weight values of each interval. Second, the depth maps predicted by feature maps with different resolutions contain different critical information. The lower-resolution depth maps play an overall role in predicting the overall contour; the higher-resolution depth map is more accurate in predicting object edge details. Finally, the desired depth map can be obtained by a 3 x 3 convolution of the predicted depth maps at multiple scales. We tested our proposed method on the NYU Depth V2 and KITTI datasets and achieved effective performance.

Automated summary

In addition to RGB information, depth information is crucial in image analysis. This paper proposes a multiscale classification network for depth estimation, which tackles the problem by transforming depth values into a nonlinear combination of multiple depth interval values. The network utilizes the correlation of depth information to determine the weight values of each interval. Furthermore, the method employs feature maps with different resolutions to predict depth maps, with lower-resolution maps capturing overall contour and higher-resolution maps providing more accurate object edge details. The desired depth map is obtained through convolution of predicted depth maps at multiple scales.