This is the implementation of our paper ``PEdger++: Practical Edge Detection via Assembling Cross Information"

Introduction

Edge detection, serving as a crucial component in numerous vision-based applications, aims to effectively extract object boundaries and/or salient edges from natural images. To be viable for broad deployment across devices with varying computational capacities, edge detectors shall balance high accuracy with low computational complexity. This paper addresses the challenge of achieving that balance: {how to efficiently capture discriminative features without relying on large-size and sophisticated models}. We propose PEdger++, a collaborative learning framework designed to reduce computational costs and model sizes while improving edge detection accuracy. The core principle of our PEdger++ is that cross-information derived from heterogeneous architectures, diverse training moments, and multiple parameter samplings, is beneficial to enhance learning from an ensemble perspective. Extensive ablation studies together with experimental comparisons on the BSDS500, NYUD and Multicue datasets demonstrate the effectiveness of our approach, both quantitatively and qualitatively, showing clear improvements over existing methods. We also provide multiple versions of the model with varying computational requirements, highlighting PEdger++'s adaptability with respect to different resource constraints.

Environment Installation

To install the packages and dependencies, run:

bash install.sh

All the pakcages can be prepared after running "install.sh".

Preparing Data

Download the augmented BSDS and PASCAL VOC datasets from:

http://mftp.mmcheng.net/liuyun/rcf/data/HED-BSDS.tar.gz

http://mftp.mmcheng.net/liuyun/rcf/data/PASCAL.tar.gz

Download the augmented NYUD dataset from:

https://pan.baidu.com/s/1J5z6235tv1xef3HXTaqnKg Extraction Code:t2ce

Training

Before starting the training process, run:

python lmdb_io.py

If you want to train our PEdger++, change the data path of training images, and then run:

python train.py

Testing

If you want to test our pre-trained model, change the data path of testing images, and then run:

python test_edge.py

Our pre-trained models are stored in the "models/" folder, and the relative path to this folder is specified in the "test_edge.py" file.

Evaluation

The matlab code for evaluation can be downloaded in https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/resources.html. Before evaluation, the non-maximum suppression should be done through running ``edge_nms.m" in https://github.com/yun-liu/RCF. The codes for plotting Precision-Recall curves are in https://github.com/yun-liu/plot-edge-pr-curves.