Swapping YOLO-World's C2f backbone blocks for attention-based A2C2f blocks raises mAP@0.5 from 32.5 to 35.2 on VisDrone, according to the paper's Table 2—which its own conclusions contradict.
Use of AI applications for the drone industry,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Attention from Above: A Multimodal Model for Drone-Based Object Localization
Swapping YOLO-World's C2f backbone blocks for attention-based A2C2f blocks raises mAP@0.5 from 32.5 to 35.2 on VisDrone, according to the paper's Table 2—which its own conclusions contradict.