Learning Relaxed Deep Supervision for Better Edge Detection

By Y. Liu, Michael S. Lew. 2016

138 citations recorded by the source corpus.

We propose using relaxed deep supervision (RDS) within convolutional neural networks for edge detection. The conventional deep supervision utilizes the general groundtruth to guide intermediate predictions. Instead, we build hierarchical supervisory signals with additional relaxed labels to consider the diversities in deep neural networks. We begin by capturing the relaxed labels from simple detectors (e.g. Canny). Then we merge them with the general groundtruth to generate the RDS. Finally we employ the RDS to supervise the edge network following a coarse-to-fine paradigm. These relaxed labels can be seen as some false positives that are difficult to be classified. Weconsider these false positives in the supervision, and are able to achieve high performance for better edge detection. Wecompensate for the lack of training images by capturing coarse edge annotations from a large dataset of image segmentations to pretrain the model. Extensive experiments demonstrate that our approach achieves state-of-the-art performance on the well-known BSDS500 dataset (ODS F-score of .792) and obtains superior cross-dataset generalization results on NYUD dataset.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Yi Yang
3. Ming-Hsuan Yang
4. Xiaogang Wang
5. Dacheng Tao
6. Qi Tian
7. Ming-Ming Cheng
8. Ling Shao

Citations in the corpus

• EDTER: Edge Detection with Transformer (2022, methodology): “Moreover, most previous works [22, 36, 37, 62] use PASCAL VOC Context Dataset [17] as the additional training data, which provides full-scene segmentation annotations with”
• Richer Convolutional Features for Edge Detection (2016, methodology): “[26] used relaxed labels generated by bottom-up edges to guide the training process of HED.”
• Deep Crisp Boundaries (2017, methodology): “Without multi-scale testing, CED already achieves better results than the top-performing method [23] in all 3 metrics.”
• Bi-Directional Cascade Network for Perceptual Edge Detection (2019, methodology): “[31] propose to relax the supervisions on intermediate layers using Canny [4] detectors with layer-specific scales.”
• RINDNet: Edge Detection for Discontinuity in Reflectance, Illumination, Normal and Depth (2021, background): “CNN-based edge detectors [3, 4, 9, 10, 17, 20, 24, 26, 31, 36, 45] do not rely on hand-crafted features and achieve better performance.”
• Pixel Difference Networks for Efficient Edge Detection (2021, methodology): “The candidate points can be also used as auxiliary relaxed labels for better training the CNN model [32].”
• HybridPose: 6D Object Pose Estimation Under Hybrid Representations (2020, background): “Edges are known to capture important image features such as object contours [2], salient edges [23], and straight line segments [45].”