Synthesizing Supervision for Learning Deep Saliency Network without Human Annotation

Computer Science

• Annotation 100%
• Art Performance 25%
• Deep Learning Method 100%
• Deep Neural Network 25%
• External Knowledge 25%
• Field Research 25%
• Learning Framework 100%
• Object Detector 100%
• Performance Gap 25%
• Process Synthesis 25%
• Salient Object Detection 25%
• Segmentation Task 25%
• Supervised Learning 25%
• Training Image 25%
• Weakly-Supervised Semantic Segmentation 50%