Search papers, labs, and topics across Lattice.
ScopeMamba-YOLO addresses the degradation of tiny object detection in remote sensing imagery by isolating state-space selective scanning into an off-path, zero-gated auxiliary stream rather than embedding it directly in the primary convolutional pipeline. This decoupling prevents Mamba operators from corrupting weak, high-resolution local features while expanding the stride-8 peripheral effective receptive field energy ratio more than tenfold (from 0.008 to 0.090). On the VisDrone-2019 benchmark, ScopeMamba-S achieves 50.8% mAP50 with only 3.57M parameters, outperforming YOLOv8s by 10.8 pp while reducing parameter count by 68%.
Forcing state-space models into the primary convolutional path actively harms tiny object detection, but routing selective scanning off-path enables an efficient detector that beats YOLOv8s by 10.8 pp on VisDrone with two-thirds fewer parameters.
Small object detection in unmanned aerial vehicle (UAV) and remote sensing imagery requires preserving high-resolution detail while modeling long-range context. Adding a stride-4 detection level and removing the stride-32 stage benefits tiny targets but weakens peripheral spatial support, whereas directly inserting selective scanning into the main feature path can interfere with weak local cues. We propose ScopeMamba-YOLO, built around an off-path, zero-gated selective-scanning principle that decouples contextual modeling from the convolutional stream. The principle is instantiated by a Cascaded Global-Context Module (CGCM) in the backbone and a Selective-Scan PAN (SS-PAN) in the neck. An Adaptive Multi-scale Strip (AMS) Block reduces the cost of high-resolution feature extraction, while a Scale-Adaptive DFL (SA-DFL) head reallocates distributional support and regression capacity across scales with only 0.008M additional parameters. Controlled experiments show that matched main-path selective scanning reduces mAP50 by 0.98 pp, whereas off-path CGCM improves the final configuration by 0.67 pp over the three-seed no-CGCM mean; operator controls indicate that this gain is not explained by auxiliary branch capacity alone. ERF analysis further shows that the complete context pathway increases the peripheral energy ratio from 0.008 to 0.090 at stride 8. On VisDrone-2019, ScopeMamba-S achieves 50.8% mAP50 with 3.57M parameters, exceeding YOLOv8s by 10.8 pp while using 32% of its parameters; ScopeMamba-M reaches 52.6% mAP50 with 6.48M parameters. Consistent improvements are also observed on AI-TOD, especially for very-tiny and tiny objects.
