Fast Camouflaged Object Detection via Edge-based Reversible Re-calibration Network

• Published in: Pattern recognition Vol. 123; p. 108414
• Main Authors: Ji, Ge-Peng, Zhu, Lei, Zhuge, Mingchen, Fu, Keren
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.03.2022
• Subjects: Camouflaged Object Detection; NGES Priors; Reversible Re-calibration Unit; Selective Edge Aggregation; Selective Edge Aggregation; Reversible Re-calibration Unit; NGES Priors; Camouflaged Object Detection
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2021.108414

•We propose a novel architecture called Edge-based Reversible Re-calibration Network (ERRNet) for camouflaged object detection. It advances the state-of-the-art performance with real-time inference speed on eight challenging datasets.•We design an aggregation strategy, Selective Edge Aggregation (SEA), to obtain initial edge prior, which can well alleviate the “ambiguous” problem of weak boundaries.•To model the cross-comparison stage of visual perception, we further design a multivariate calibration strategy, termed Reversible Re-calibration Unit (RRU), which re-calibrates the coarse inference map by considering diverse priors. Camouflaged Object Detection (COD) aims to detect objects with similar patterns (e.g., texture, intensity, colour, etc) to their surroundings, and recently has attracted growing research interest. As camouflaged objects often present very ambiguous boundaries, how to determine object locations as well as their weak boundaries is challenging and also the key to this task. Inspired by the biological visual perception process when a human observer discovers camouflaged objects, this paper proposes a novel edge-based reversible re-calibration network called ERRNet. Our model is characterized by two innovative designs, namely Selective Edge Aggregation (SEA) and Reversible Re-calibration Unit (RRU), which aim to model the visual perception behaviour and achieve effective edge prior and cross-comparison between potential camouflaged regions and background. More importantly, RRU incorporates diverse priors with more comprehensive information comparing to existing COD models. Experimental results show that ERRNet outperforms existing cutting-edge baselines on three COD datasets and five medical image segmentation datasets. Especially, compared with the existing top-1 model SINet, ERRNet significantly improves the performance by ∼6% (mean E-measure) with notably high speed (79.3 FPS), showing that ERRNet could be a general and robust solution for the COD task.