Depth from Monocular Images using a Semi-Parallel Deep Neural Network (SPDNN) Hybrid Architecture
share
Convolutional Neural Network (CNN) techniques are applied to the problem of determining the depth from a single camera image (monocular depth). Fully connected CNN topologies preserve all details of the input images, enabling the detection of fine details, but miss larger features; networks that employ 2x2, 4x4 and 8x8 max-pooling operators can determine larger features at the expense of finer details. After designing, training and optimising a set of topologies, these networks may be combined into a single network topology using graph optimization techniques. This "Semi Parallel Deep Neural Network (SPDNN)" eliminates duplicate common network layers, reducing network size and computational effort significantly, and can be further optimized by retraining to achieve an improved level of convergence over the individual topologies. In this study, four models are trained and have been evaluated at 2 stages on the KITTI dataset. The ground truth images in the first part of the experiment come from the benchmark, and for the second part, the ground truth images are the depth map results from applying a state-of-the-art stereo matching method. The results of this evaluation demonstrate that using post-processing techniques to refine the target of the network increases the accuracy of depth estimation on individual mono images. The second evaluation shows that using segmentation data as the input can improve the depth estimation results to a point where performance is comparable with stereo depth estimation. The computational time is also discussed in this study.
READ FULL TEXT