Reproduce “YOLOF-MLD: A Multi-Level Distillation Framework for Efficient Object Detection” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiaolong Wu, Liming Zheng, Zitong Liu, Yue Zheng
YOLOF-MLD trains a single-input/single-output detector for fire and smoke using an FPN teacher. It distills backbone features, neck features, and predictions into a simpler SiSo student, which runs without the teacher at inference. On D-Fire and AFS, it uses 1.53M parameters and reaches 12.76 FPS on Raspberry Pi 5, with lower D-Fire mAP50 than YOLO26n.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.