Depth Adaptive Deep Neural Network for Semantic Segmentation

TL;DR

This paper introduces a depth-adaptive neural network with a novel convolution layer that dynamically adjusts receptive fields based on depth information, improving semantic segmentation accuracy across varying object distances.

Contribution

The paper proposes the depth-adaptive multiscale (DaM) convolution layer, enabling neural networks to adapt receptive fields at each neuron using depth data, enhancing segmentation performance.

Findings

Outperforms state-of-the-art methods on RGB-D datasets

Effective in hand-object interaction segmentation

No additional layers or pre/post-processing needed

Abstract

In this work, we present the depth-adaptive deep neural network using a depth map for semantic segmentation. Typical deep neural networks receive inputs at the predetermined locations regardless of the distance from the camera. This fixed receptive field presents a challenge to generalize the features of objects at various distances in neural networks. Specifically, the predetermined receptive fields are too small at a short distance, and vice versa. To overcome this challenge, we develop a neural network which is able to adapt the receptive field not only for each layer but also for each neuron at the spatial location. To adjust the receptive field, we propose the depth-adaptive multiscale (DaM) convolution layer consisting of the adaptive perception neuron and the in-layer multiscale neuron. The adaptive perception neuron is to adjust the receptive field at each spatial location using the corresponding depth information. The in-layer multiscale neuron is to apply the different size of the receptive field at each feature space to learn features at multiple scales. The proposed DaM convolution is applied to two fully convolutional neural networks. We demonstrate the effectiveness of the proposed neural networks on the publicly available RGB-D dataset for semantic segmentation and the novel hand segmentation dataset for hand-object interaction. The experimental results show that the proposed method outperforms the state-of-the-art methods without any additional layers or pre/post-processing.