Beyond Universal Saliency: Personalized Saliency Prediction with Multi-task CNN
Authors: Yanyu Xu, Nianyi Li, Junru Wu, Jingyi Yu, Shenghua Gao
IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comprehensive experiments demonstrate that our new PSM model and prediction scheme are effective and reliable. and 5 Experiments 5.1 Experimental Setup Parameters. We implement our solution on the CAFFE framework [Jia et al., 2014]. and The performance of all methods are listed in Table 2. |
| Researcher Affiliation | Collaboration | Yanyu Xu1 , Nianyi Li2,3, Junru Wu1, Jingyi Yu1,3, and Shenghua Gao1 1Shanghai Tech University, Shanghai, China. 2University of Delaware, Newark, DE, USA. 3 Plex-VR digital technology Co., Ltd. |
| Pseudocode | No | The paper describes the architecture of the Multi-task CNN in text and via Figure 4, but it does not include any labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any specific link or statement indicating that the source code for their method is publicly available. |
| Open Datasets | Yes | To do so, we first analyze existing datasets. ... 1,100 images are chosen from existing saliency detection datasets including SALICON [Jiang et al., 2015], Image Net [Russakovsky et al., 2015], i SUN [Xu et al., 2015], OSIE[Xu et al., 2014], PASCAL-S [Li et al., 2014] |
| Dataset Splits | No | In our experiments, we randomly select 600 images ar training data, and use the rest 1,000 images for testing. The paper does not explicitly mention a separate validation split or its size. |
| Hardware Specification | No | The paper states 'We implement our solution on the CAFFE framework [Jia et al., 2014]' but does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used for conducting the experiments. |
| Software Dependencies | No | The paper mentions implementing the solution on 'the CAFFE framework [Jia et al., 2014]' but does not specify any version numbers for CAFFE or other software dependencies. |
| Experiment Setup | Yes | We train our network with the following hyper-parameters setting: mini-batch size (40), learning rate (0.0003), momentum (0.9), weight decay (0.0005), and number of iterations (40,000). |