AARN: Anchor‐guided attention refinement network for inshore ship detection

Inshore ship detection is an important task in several fields, for example, maritime transportation, supervision, and port management. However, due to the diversified categories locations of different ships interference complex surroundings, capturing discriminative characteristics multi-scale inshore accurate still challenging. Here, anchor-guided attention refinement network (AARN) proposed alleviate problems by prominently designing feature filter module (AFFM) alignment (AADM). In AFFM, which supervision generated from high-level semantic features, used highlight informative target cues suppress background when establishing a four-layer pyramid. AADM, anchor-aligned features are adopted eventually identify potential ships, both alleviates misalignment between refined anchors pyramidal improves performance further. Extensive experiments conducted on public Seaships7000 dataset verify contributions modules effectiveness our method detecting comparison domain-specific general CNN-based methods.