联合正交特征融合与大核可分离注意力的道路分割算法

汤毅杰,钟铭恩,袁彬淦,范康,谭佳威,林志强

Road segmentation algorithm based on joint orthogonal feature fusion and large kernel separable attention

Yijie TANG,Mingen ZHONG,Bingan YUAN,Kang FAN,Jiawei TAN,Zhiqiang LIN

表 3 不同剪枝方法性能对比结果

Tab.3 Performance comparison of different pruning methods

方法 Massachusetts数据集 DeepGlobe数据集 参数量/106 FLOPs/109 P/% R/% IoU/% APLS/% P/% R/% IoU/% APLS/% 未剪枝 78.65 79.98 65.12 75.36 80.21 81.98 68.41 75.93 37.3 29.2 L1[29] 74.62 75.88 61.47 71.83 76.35 77.68 64.58 72.78 22.5 14.9 Group norm[30] 75.58 76.93 62.38 72.08 77.63 78.74 65.47 73.51 22.8 15.3 Group hessian[31] 74.96 76.39 61.95 71.92 76.53 78.31 64.89 72.87 22.7 15.4 Group taylor[32] 74.35 75.43 61.18 71.69 76.02 77.33 64.08 72.28 22.8 15.1 LAMP 78.12 79.63 64.85 74.95 79.78 81.63 67.88 75.53 22.5 14.8