Object Contour Detection with a Fully Convolutional Encoder-Decoder Network

Abstract

We develop a deep learning algorithm for contour detection with a fully convolutional encoder-decoder network. Different from previous low-level edge detection, our algorithm focuses on detecting higher-level object contours. Our network is trained end-to-end on PASCAL VOC with refined ground truth from inaccurate polygon annotations, yielding much higher precision in object contour detection than previous methods. We find that the learned model generalizes well to unseen object classes from the same super-categories on MS COCO and can match state-of-the-art edge detection on BSDS500 with fine-tuning. By combining with the multiscale combinatorial grouping algorithm, our method can generate high-quality segmented object proposals, which significantly advance the state-of-the-art on PASCAL VOC (improving average recall from 0.62 to 0.67) with a relatively small amount of candidates ($\sim$1660 per image).

Accepted by CVPR2016 as spotlight

Cited by in corpus (19)

• BoxCars: Improving Fine-Grained Recognition of Vehicles using 3-D Bounding Boxes in Traffic Surveillance
• Traffic Surveillance Camera Calibration by 3D Model Bounding Box Alignment for Accurate Vehicle Speed Measurement
• Learning Deep Structured Multi-Scale Features using Attention-Gated CRFs for Contour Prediction
• Generative Face Completion
• Learning a Discriminative Feature Network for Semantic Segmentation
• Faster Training of Mask R-CNN by Focusing on Instance Boundaries
• CASENet: Deep Category-Aware Semantic Edge Detection
• Weakly-supervised Salient Instance Detection
• Recent Advances in the Applications of Convolutional Neural Networks to Medical Image Contour Detection
• Deep Motion Boundary Detection
• Contextual Guided Segmentation Framework for Semi-supervised Video Instance Segmentation
• Instance-Level Salient Object Segmentation
• Distance to Center of Mass Encoding for Instance Segmentation
• Hi-Fi: Hierarchical Feature Integration for Skeleton Detection
• DOOBNet: Deep Object Occlusion Boundary Detection from an Image
• Pixel-wise object tracking
• DDNet: Dual-path Decoder Network for Occlusion Relationship Reasoning
• Human Recognition Using Face in Computed Tomography
• SRN: Side-output Residual Network for Object Reflection Symmetry Detection and Beyond