An Improved Object Detection Algorithm Based on M2Det
Tao Zhang, Linwei Li · 2020
This paper proposes an improved target detection algorithm based on M2Det, which improves the way to construct feature pyramids in M2Det, and improves the detection accuracy on the Apple dataset by stacking MLFPN modules. The experimental results prove that the two-layer MLFPN module achieves the best performance. Compared with the original M2Det model, [email protected] and [email protected] have improved by 11.8 and 10.5, respectively.