Saliency Detection with Recurrent Fully Convolutional Networks
View Researcher's Other CodesDisclaimer: The provided code links for this paper are external links. Science Nest has no responsibility for the accuracy, legality or content of these links. Also, by downloading this code(s), you agree to comply with the terms of use as set out by the author(s) of the code(s).
Authors | Linzhao Wang, Lijun Wang, Huchuan Lu, Pingping Zhang, Xiang Ruan |
Journal/Conference Name | ECCV |
Paper Category | ECE |
Paper Abstract | • Employs three kind of low-level contrast features, including color, intensity and orientation, and the center prior knowledge to introduce saliency prior maps. • Train the RFCN with two stage training strategy, pre-training on the segmentation data set and fine-tuning on the saliency data set. The recurrent structure can incorporate the saliency prior maps into the CNNs with an end-to-end training method. • Refine the saliency maps with a postprocessing method which can improve the performance of RFCN. Edge-preserving maps are produced with the computation of color confidence and spatial confidence. 3. Algorithm 4. Results |
Date of publication | 2016 |
Code Programming Language | MATLAB |
Comment |