最新版本的NCNN
模型 运行optimise 时报错
NOTE: The input of darknet uses: mean_vals=0 and norm_vals=1/255.f.
NOTE: Remeber to use ncnnoptimize for better performance.
fuse_convolution_batchnorm 0_26 0_26_bn
fuse_convolution_batchnorm 2_38 2_38_bn
fuse_convolution_batchnorm 4_50 4_50_bn
fuse_convolution_batchnorm 6_62 6_62_bn
fuse_convolution_batchnorm 8_74 8_74_bn
fuse_convolution_batchnorm 10_86 10_86_bn
fuse_convolution_batchnorm 12_98 12_98_bn
fuse_convolution_batchnorm 13_108 13_108_bn
fuse_convolution_batchnorm 14_116 14_116_bn
fuse_convolution_batchnorm 18_146 18_146_bn
fuse_convolution_batchnorm 21_160 21_160_bn
fuse_convolution_batchnorm 25_190 25_190_bn
fuse_convolution_batchnorm 28_204 28_204_bn
fuse_convolution_activation 0_26 0_26_bn_leaky
fuse_convolution_activation 2_38 2_38_bn_leaky
fuse_convolution_activation 4_50 4_50_bn_leaky
fuse_convolution_activation 6_62 6_62_bn_leaky
fuse_convolution_activation 8_74 8_74_bn_leaky
fuse_convolution_activation 10_86 10_86_bn_leaky
fuse_convolution_activation 12_98 12_98_bn_leaky
fuse_convolution_activation 13_108 13_108_bn_leaky
fuse_convolution_activation 14_116 14_116_bn_leaky
fuse_convolution_activation 18_146 18_146_bn_leaky
fuse_convolution_activation 21_160 21_160_bn_leaky
fuse_convolution_activation 25_190 25_190_bn_leaky
fuse_convolution_activation 28_204 28_204_bn_leaky
shape_inference
Segmentation fault (core dumped)
1条答案
按热度按时间qzlgjiam1#
这一层 yolo 的maxpool 转ncnn 后会从 19X19 变到 18X18 应该怎么调整比较好？