Dense Extreme Inception Network for Edge Detection
This work addresses the lack of curated datasets for edge detection, which is crucial for computer vision applications, though it is incremental in improving dataset quality and network design.
The authors tackled the problem of edge detection in computer vision by introducing a new dataset that distinguishes edges, contours, and boundaries, and proposing DexiNed, a novel architecture that outperforms other algorithms on this dataset and generalizes well without fine-tuning.
<<<This is a pre-acceptance version, please, go through Pattern Recognition Journal on Sciencedirect to read the final version>>>. Edge detection is the basis of many computer vision applications. State of the art predominantly relies on deep learning with two decisive factors: dataset content and network's architecture. Most of the publicly available datasets are not curated for edge detection tasks. Here, we offer a solution to this constraint. First, we argue that edges, contours and boundaries, despite their overlaps, are three distinct visual features requiring separate benchmark datasets. To this end, we present a new dataset of edges. Second, we propose a novel architecture, termed Dense Extreme Inception Network for Edge Detection (DexiNed), that can be trained from scratch without any pre-trained weights. DexiNed outperforms other algorithms in the presented dataset. It also generalizes well to other datasets without any fine-tuning. The higher quality of DexiNed is also perceptually evident thanks to the sharper and finer edges it outputs.