Abstract: Camouflaged object detection （COD）， which focuses on identifying and locating camouflaged objects within images， is a crucial task in the computer-vision community. This task is essential in various fields， such as medical imaging and wildlife monitoring. Although COD algorithms have been developed significantly， it cannot effectively detect small camouflaged objects， which typically occupy less than 10% of the image area. Hence， We propose a novel approach CAHNet， namely context-aware and hierarchical network， to precisely detect small camouflaged objects. Specifically， we designed a semantic-guided hierarchical cross-scale fusion module to integrate cross-scale contextual information during the encoding stage， which results in better feature embedding. Furthermore， we developed a patch-integrated hierarchical decoding module to effectively capture spatially contextual relationships across global and local contexts. This enables the CAHNet to acquire a broader contextual sensing ability， thus resulting in better detection performance. Additionally， we developed four new testbeds， i.e.， CHAMELEON-ts， CAMO-ts， COD10K-ts， and NC4K-ts， to evaluate the detection performance on small camouflaged objects. Extensive experimental results on these benchmarks clearly show that the CAHNet achieved advanced detection results on small camouflaged objects， thus outperforming existing state-of-the-art methods. In summary， the CAHNet provides a robust solution for applications requiring precise and reliable detections of small camouflaged objects in complex environments.