Plant Camouflage Detection: A New Dataset and Benchmark
This paper introduces the novel problem of plant camouflage detection (PCD) and presents a new dataset, PlantCamo, for benchmarking PCD algorithms. The authors find that existing camouflaged object detection (COD) models perform poorly on PlantCamo due to the unique characteristics of plant camouflage, and propose a new model, PCNet, which achieves superior performance on this task.
Mamba Capsule Routing for Camouflaged Object Detection: A Part-Whole Relational Approach
This paper introduces Mamba Capsule Routing Network (MCRNet), a novel approach for camouflaged object detection that leverages the part-whole relational properties of Capsule Networks and the efficiency of Vision Mamba for lightweight capsule routing, achieving state-of-the-art performance on three benchmark datasets.
Hierarchical Graph Interaction Transformer with Dynamic Token Clustering for Effective Camouflaged Object Detection
The proposed HGINet model can effectively discover imperceptible camouflaged objects by employing a hierarchical graph interaction mechanism and dynamic token clustering strategy.
Frequency-Guided Spatial Adaptation for Enhancing Camouflaged Object Detection
A novel frequency-guided spatial adaptation method is proposed to enhance the feature representation for accurately detecting camouflaged objects by dynamically adjusting the frequency components to focus more on the concealed regions.
Collaborative Optimization Strategy for Effective Camouflaged Object Detection
A novel collaborative optimization strategy that simultaneously models long-range dependencies and local details to generate high-quality features for accurate detection of camouflaged objects.
Adaptive Guidance Learning Network for Accurate Camouflaged Object Detection
The proposed adaptive guidance learning network (AGLNet) can effectively explore and integrate various additional cues, such as boundary, texture, edge, and frequency information, to guide the learning of camouflaged features for accurate object detection.
Efficient Single-branch Network for Camouflaged Object Detection via Spotlight Shifting Co-supervision
A novel single-branch network, Co-Supervised Spotlight Shifting Network (CS3Net), efficiently leverages a spotlight shifting strategy for co-supervision to enhance the detection of camouflaged objects.
Strategische Beute machen scharfe Raubtiere: Verbesserung von getarnten Objektdetektoren durch die Erzeugung getarnter Objekte
Algorithmen zur Verbesserung von getarnten Objektdetektoren durch die Erzeugung getarnter Objekte.
Preys and Predators in Camouflaged Object Detection
Preys develop better camouflage, predators acquire acute vision.
Preys and Predators in Camouflaged Object Detection: A Novel Approach
Enhancing camouflaged object detectors through adversarial training and feature coherence.
© 2024 by Linnk AI