To improve the accuracy of lightweight detectors, researchers have extensively employed feature-level knowledge distillation. However, existing methods often rely on ground-truth bounding boxes to localize salient regions, thereby directing the representation learning of student models. These approaches ignore the significant structural and semantic divergence of teacher–student networks, making it challenging to transfer effective knowledge. Our method, BAD, constitutes a new framework for distillation that introduces bidirectional attention guidance across spatial and channel dimensions. The core of BAD is to extract spatial response patterns from both teacher and student networks and fuse them into a joint attention mask. This mask identifies semantically aligned regions without relying on ground-truth boxes, enabling adaptive guidance for the student to focus on task-relevant areas while preserving its representational flexibility. Meanwhile, the masking mechanism helps mitigate training imbalance caused by overly strict alignment. The spatial attention module helps localize contextually important regions, while the channel attention enhances global semantic alignment. We perform comprehensive experiments on multiple popular object detectors, including YOLOv8, RetinaNet, FCOS, and Faster R-CNN. The results demonstrate that BAD achieves stable and effective knowledge transfer under heterogeneous architectures and limited model capacity.