Spatial Information Guided Convolution for Real-Time RGBD Semantic Segmentation
This work addresses the need for real-time RGBD semantic segmentation in applications like robotics or augmented reality, offering an incremental improvement over two-stream methods.
The paper tackles the problem of inefficient integration of 3D spatial information in RGBD semantic segmentation by proposing Spatial information guided Convolution (S-Conv), which achieves real-time inference and state-of-the-art performance on NYUDv2 and SUNRGBD datasets.
3D spatial information is known to be beneficial to the semantic segmentation task. Most existing methods take 3D spatial data as an additional input, leading to a two-stream segmentation network that processes RGB and 3D spatial information separately. This solution greatly increases the inference time and severely limits its scope for real-time applications. To solve this problem, we propose Spatial information guided Convolution (S-Conv), which allows efficient RGB feature and 3D spatial information integration. S-Conv is competent to infer the sampling offset of the convolution kernel guided by the 3D spatial information, helping the convolutional layer adjust the receptive field and adapt to geometric transformations. S-Conv also incorporates geometric information into the feature learning process by generating spatially adaptive convolutional weights. The capability of perceiving geometry is largely enhanced without much affecting the amount of parameters and computational cost. We further embed S-Conv into a semantic segmentation network, called Spatial information Guided convolutional Network (SGNet), resulting in real-time inference and state-of-the-art performance on NYUDv2 and SUNRGBD datasets.