In modern internet of things (IoT), visual analysis and predictions are often performed by deep learning models. Salient object detection (SOD) is a fundamental pre-processing for these applications. Executing SOD on the fog devices is a challenging task due to the diversity of data and fog devices. To adopt convolutional neural networks (CNN) on fog-cloud infrastructures for SOD-based applications, we introduce a semisupervised adversarial learning method in this paper. The proposed model, named as SaliencyGAN, is empowered by a novel concatenated-GAN framework with partially shared parameters. The backbone CNN can be chosen flexibly based on the specific devices and applications. In the meanwhile, our method uses both the labelled and unlabelled data from different problem domains for training. Using multiple popular benchmark datasets, we compared state-of-the-art baseline methods to our SaliencyGAN obtained with 10% to 100% labelled training data. SaliencyGAN gained performance comparable to the supervised baselines when the percentage of labelled data reached 30%, and outperformed the weakly supervised and unsupervised baselines. Furthermore, our ablation study shows that SaliencyGAN were more more robust to the common "mode missing" (or "mode collapse") issue compared to the selected popular GAN models. The visualized ablation results proved that SaliencyGAN learned a better estimation of data distributions. To the best of our knowledge, this is the first IoT-oriented semi-supervised SOD method.

No abstract

No abstract

No abstract