Contrast Prior and Fluid Pyramid Integration for RGBD Salient Object Detection

The large availability of depth sensors provides valuable complementary information for salient object detection (SOD) in RGBD images. However, due to the inherent difference between RGB and depth information, extracting features from the depth channel using ImageNet pre-trained backbone models and fusing them with RGB features directly are sub-optimal. In this paper, we utilize contrast prior, which used to be a dominant cue in none deep learning based SOD approaches, into CNNs-based architecture to enhance the depth information. The enhanced depth cues are further integrated with RGB features for SOD, using a novel fluid pyramid integration, which can make better use of multi-scale cross-modal features. Comprehensive experiments on 5 challenging benchmark datasets demonstrate the superiority of the architecture CPFP over 9 state-of-the-art alternative methods.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Benchmark
RGB-D Salient Object Detection DES CPFP S-Measure 87.2 # 11
Average MAE 0.038 # 11
max E-Measure 92.3 # 7
max F-Measure 84.6 # 8
RGB-D Salient Object Detection LFSD CPFP S-Measure 82.8 # 6
Average MAE 0.088 # 6
max E-Measure 86.3 # 3
max F-Measure 81.1 # 3
RGB-D Salient Object Detection NJU2K CPFP S-Measure 87.8 # 22
Average MAE 0.053 # 22
max E-Measure 92.6 # 10
max F-Measure 87.7 # 12
RGB-D Salient Object Detection NLPR CPFP S-Measure 88.8 # 12
Average MAE 0.036 # 12
max F-Measure 86.7 # 9
max E-Measure 93.2 # 8
RGB-D Salient Object Detection RGBD135 CPFP S-Measure 80.7 # 5
Average MAE 0.082 # 5
max F-Measure 76.6 # 5
max E-Measure 85.2 # 4
RGB-D Salient Object Detection SIP CPFP S-Measure 85.0 # 12
max E-Measure 90.3 # 8
max F-Measure 85.1 # 8
Average MAE 0.064 # 13
RGB-D Salient Object Detection STERE CPFP S-Measure 87.9 # 12
Average MAE 0.051 # 12
max F-Measure 87.4 # 9
max E-Measure 92.5 # 8

Methods


No methods listed for this paper. Add relevant methods here