TY - JOUR
T1 - LTDNet: A Lightweight Two-Stage Decoder Network for RGB-D Salient Object Detection
AU - Wang , Jian
AU - Chen , Wenbing
JO - Journal of Information and Computing Science
VL - 2
SP - 102
EP - 117
PY - 2024
DA - 2024/01
SN - 17
DO - http://doi.org/
UR - https://global-sci.org/intro/article_detail/jics/22353.html
KW - salient object detection, RGB-D, lightweight, efficient.
AB - <p style="text-align: justify;">Most existing models of RGB-D salient object detection (SOD) utilize heavy backbones like
VGGs and ResNets which lead to large model size and high computational costs. In order to improve this
problem, a lightweight two-stage decoder network is proposed. Firstly, the network utilizes MobileNet-V2 and
a customized backbone to extract the features of RGB images and depth maps respectively. In order to mine
and combine cross-modality information, cross reference module is used to fuse complementary information
from different modalities. Subsequently, we design a feature enhancement module to enhance the clues of the
fused features which has four parallel convolutions with different expansion rates. Finally, a two-stage decoder
is used to predict the saliency maps, which processes high-level features and low-level features separately and
then merges them. Experiments on 5 benchmark datasets comparing with 10 state-of-the-art models
demonstrate that our model can achieve significant improvement with smallest model size.</p>