YOLO-EMAC的整体结构如下图所示,可以看出,整体结构就是YOLOv12,把其中的一些模块替换成了M2C2f和A2C2f.
# YOLO-EMAC 🚀 (based on YOLOv12-turbo) # Replaces A2C2f with M2C2f and C3k2 with C3k2_EAMC (E-C3k2) # Parameters nc: 80 # number of classes scales: # model compound scaling constants n: [0.50, 0.25, 1024] s: [0.50, 0.50, 1024] m: [0.50, 1.00, 512] l: [1.00, 1.00, 512] x: [1.00, 1.50, 512] # YOLO-EMAC backbone backbone: - [-1, 1, Conv, [64, 3, 2]] # 0-P1/2 - [-1, 1, Conv, [128, 3, 2, 1, 2]] # 1-P2/4 - [-1, 2, C3k2_EAMC, [256, False, 0.25, 1, False]] # 2 (E-C3k2) - [-1, 1, Conv, [256, 3, 2, 1, 4]] # 3-P3/8 - [-1, 2, C3k2_EAMC, [512, False, 0.25, 1, False]] # 4 (E-C3k2) - [-1, 1, Conv, [512, 3, 2]] # 5-P4/16 - [-1, 4, M2C2f, [512]] # 6 (M2C2f) - [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32 - [-1, 4, M2C2f, [1024]] # 8 (M2C2f) # YOLO-EMAC head head: - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 6], 1, Concat, [1]] # cat backbone P4 - [-1, 2, M2C2f, [512]] # 11 (M2C2f) - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 4], 1, Concat, [1]] # cat backbone P3 - [-1, 2, C3k2_EAMC, [256, False, 0.5, 1, False]] # 14 (E-C3k2) - [-1, 1, Conv, [256, 3, 2]] - [[-1, 11], 1, Concat, [1]] # cat head P4 - [-1, 2, M2C2f, [512]] # 17 (M2C2f) - [-1, 1, Conv, [512, 3, 2]] - [[-1, 8], 1, Concat, [1]] # cat head P5 - [-1, 2, C3k2_EAMC, [1024, False, 0.5, 1, True]] # 20 (E-C3k2) - [[14, 17, 20], 1, Detect, [nc]] # Detect(P3, P4, P5)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。