if ( WINRT )
ocv_module_disable ( dnn )
endif ( )
set ( the_description "Deep neural network module. It allows to load models from different frameworks and to make forward pass" )
ocv_add_dispatched_file_force_all ( "layers/layers_common" AVX AVX2 AVX512_SKX RVV LASX )
ocv_add_dispatched_file_force_all ( "int8layers/layers_common" AVX2 AVX512_SKX RVV LASX )
ocv_add_dispatched_file_force_all ( "layers/cpu_kernels/conv_block" AVX AVX2 NEON NEON_FP16 )
Merge pull request #23192 from zihaomu:clean_up_SIMD_code
### Purpose of this PR:
- Move all dispatch and SIMD code of `convolution layer` into `simd.hpp` file.
- Support Winograd at AVX-only machine.
- Re-name the folder from `fast_conv` to `cpu_kernels`. In the future, we can put other layers of CPU optimization into it, like `GEMM` or `MatMul`.
## Performance Test
Since this patch just focuses on the code style, the performance is expected as the same as before.
Test with the following script:
`./bin/opencv_perf_dnn '--gtest_filter=*conv*' --gtest_output="xml:../1-0th.xml" --perf_threads=1`
### Test on X86 platform
Min (ms)
|Name of Test|4.x | patch | 4.x vs patch (x-factor)|
|---|:-:|:-:|:-:|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 2, 19}, OCN=2, G=2, S=2, P=(1, 1), BIAS, OCV/CPU)|0.001|0.001|0.98|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 2, 25}, OCN=2, G=2, P=(2, 2), PM=SAME, OCV/CPU)|0.001|0.001|0.95|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 6, 10}, OCN=6, PM=VALID, BIAS, OCV/CPU)|0.001|0.001|0.97|
|conv3d::Conv3D::(GFLOPS=0.000, K=[1 x 1 x 1], IN={1, 4, 9, 10, 10}, OCN=4, S=[1 x 1 x 2], P=(1, 1) x (1, 1) x (1, 1), PM=VALID, OCV/CPU)|0.002|0.002|1.04|
|conv3d::Conv3D::(GFLOPS=0.000, K=[1 x 1 x 1], IN={1, 8, 1, 10, 10}, OCN=8, G=8, P=(1, 1) x (1, 1) x (1, 1), BIAS, OCV/CPU)|0.002|0.002|0.94|
|conv3d::Conv3D::(GFLOPS=0.000, K=[3 x 3 x 3], IN={1, 2, 19, 19, 19}, OCN=2, G=2, S=[2 x 2 x 2], P=(1, 1) x (1, 1) x (1, 1), BIAS, OCV/CPU)|0.040|0.044|0.93|
|conv3d::Conv3D::(GFLOPS=0.000, K=[3 x 4 x 2], IN={1, 4, 8, 10, 10}, OCN=4, G=4, S=[1 x 2 x 1], BIAS, OCV/CPU)|0.010|0.010|1.00|
|conv3d::Conv3D::(GFLOPS=0.001, K=[3 x 3 x 3], IN={1, 2, 25, 19, 19}, OCN=2, G=2, S=[1 x 2 x 2], P=(2, 2) x (2, 2) x (2, 2), PM=SAME, OCV/CPU)|0.106|0.103|1.03|
|conv3d::Conv3D::(GFLOPS=0.002, K=[3 x 1 x 4], IN={1, 14, 5, 10, 10}, OCN=14, PM=SAME, OCV/CPU)|0.041|0.040|1.03|
|conv3d::Conv3D::(GFLOPS=0.006, K=[5 x 5 x 5], IN={1, 4, 50, 19, 19}, OCN=4, S=[2 x 2 x 2], P=(1, 1) x (1, 1) x (1, 1), PM=VALID, OCV/CPU)|0.340|0.329|1.03|
|conv3d::Conv3D::(GFLOPS=0.027, K=[3 x 3 x 3], IN={1, 6, 10, 38, 50}, OCN=6, PM=VALID, BIAS, OCV/CPU)|0.590|0.567|1.04|
|conv3d::Conv3D::(GFLOPS=0.030, K=[5 x 5 x 5], IN={1, 6, 19, 19, 19}, OCN=6, G=2, OCV/CPU)|1.374|1.314|1.05|
|conv3d::Conv3D::(GFLOPS=0.045, K=[7 x 7 x 7], IN={1, 2, 38, 38, 38}, OCN=2, S=[1 x 2 x 1], OCV/CPU)|3.715|3.528|1.05|
|conv3d::Conv3D::(GFLOPS=0.053, K=[3 x 3 x 3], IN={1, 10, 98, 10, 10}, OCN=10, PM=SAME, OCV/CPU)|1.181|1.166|1.01|
|conv3d::Conv3D::(GFLOPS=0.071, K=[7 x 7 x 7], IN={1, 6, 15, 19, 19}, OCN=6, S=[2 x 1 x 1], P=(3, 3) x (3, 3) x (3, 3), PM=SAME, BIAS, OCV/CPU)|2.689|2.587|1.04|
|conv3d::Conv3D::(GFLOPS=0.093, K=[5 x 5 x 5], IN={1, 4, 40, 75, 75}, OCN=4, S=[2 x 2 x 2], OCV/CPU)|4.754|4.500|1.06|
|conv3d::Conv3D::(GFLOPS=0.116, K=[5 x 5 x 5], IN={1, 2, 21, 75, 100}, OCN=2, BIAS, OCV/CPU)|9.612|9.112|1.05|
|conv3d::Conv3D::(GFLOPS=1.267, K=[5 x 5 x 5], IN={1, 3, 75, 75, 100}, OCN=3, PM=SAME, BIAS, OCV/CPU)|69.000|64.676|1.07|
|conv3d::Conv3D::(GFLOPS=1.343, K=[3 x 3 x 3], IN={1, 11, 9, 150, 200}, OCN=11, PM=VALID, BIAS, OCV/CPU)|20.248|18.451|1.10|
|conv::Conv::(GFLOPS=0.177, K=[1 x 1], IN={1, 512, 26, 26}, OCN=256, OCV/CPU)|1.395|1.392|1.00|
|conv::Conv::(GFLOPS=0.177, K=[1 x 1], IN={1, 1024, 13, 13}, OCN=512, OCV/CPU)|1.990|1.984|1.00|
|conv::Conv::(GFLOPS=0.178, K=[1 x 1], IN={1, 256, 52, 52}, OCN=128, OCV/CPU)|1.393|1.360|1.02|
|conv::Conv::(GFLOPS=0.210, K=[1 x 1], IN={1, 576, 38, 50}, OCN=96, PM=SAME, BIAS, OCV/CPU)|1.813|1.744|1.04|
|conv::Conv::(GFLOPS=0.231, K=[3 x 3], IN={1, 128, 56, 56}, OCN=32, P=[1 x 1], OCV/CPU)|1.190|1.191|1.00|
|conv::Conv::(GFLOPS=0.231, K=[3 x 3], IN={1, 256, 14, 14}, OCN=256, P=[1 x 1], OCV/CPU)|1.286|1.284|1.00|
|conv::Conv::(GFLOPS=0.280, K=[1 x 1], IN={1, 576, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|2.295|2.279|1.01|
|conv::Conv::(GFLOPS=0.302, K=[3 x 3], IN={1, 64, 64, 64}, OCN=64, PM=SAME, OCV/CPU)|1.322|1.331|0.99|
|conv::Conv::(GFLOPS=0.357, K=[1 x 1], IN={1, 64, 208, 208}, OCN=64, OCV/CPU)|3.784|3.533|1.07|
|conv::Conv::(GFLOPS=0.420, K=[3 x 3], IN={1, 96, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|1.838|1.844|1.00|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 128, 40, 40}, OCN=128, PM=SAME, OCV/CPU)|1.957|1.959|1.00|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 256, 20, 20}, OCN=256, PM=SAME, OCV/CPU)|2.596|2.573|1.01|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 512, 10, 10}, OCN=512, PM=SAME, OCV/CPU)|4.183|4.083|1.02|
|conv::Conv::(GFLOPS=0.561, K=[3 x 3], IN={1, 128, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|2.413|2.406|1.00|
|conv::Conv::(GFLOPS=0.624, K=[3 x 3], IN={1, 128, 46, 46}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|2.538|2.546|1.00|
|conv::Conv::(GFLOPS=0.701, K=[3 x 3], IN={1, 128, 38, 50}, OCN=160, PM=SAME, BIAS, OCV/CPU)|2.972|2.980|1.00|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 64, 104, 104}, OCN=64, P=[1 x 1], OCV/CPU)|3.452|3.464|1.00|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 128, 52, 52}, OCN=128, P=[1 x 1], OCV/CPU)|3.082|3.105|0.99|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 256, 26, 26}, OCN=256, P=[1 x 1], OCV/CPU)|4.043|3.919|1.03|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 512, 13, 13}, OCN=512, P=[1 x 1], OCV/CPU)|5.538|5.531|1.00|
|conv::Conv::(GFLOPS=0.830, K=[3 x 3], IN={1, 64, 75, 100}, OCN=96, PM=SAME, BIAS, OCV/CPU)|3.393|3.418|0.99|
|conv::Conv::(GFLOPS=0.958, K=[3 x 3], IN={1, 192, 38, 38}, OCN=192, PM=SAME, OCV/CPU)|4.325|4.234|1.02|
|conv::Conv::(GFLOPS=0.958, K=[3 x 3], IN={1, 384, 19, 19}, OCN=384, PM=SAME, OCV/CPU)|6.009|5.908|1.02|
|conv::Conv::(GFLOPS=1.022, K=[3 x 3], IN={1, 576, 19, 19}, OCN=273, PM=SAME, BIAS, OCV/CPU)|6.557|6.376|1.03|
|conv::Conv::(GFLOPS=1.112, K=[3 x 3], IN={1, 512, 10, 10}, OCN=1206, P=[1 x 1], BIAS, OCV/CPU)|10.114|9.472|1.07|
|conv::Conv::(GFLOPS=1.181, K=[3 x 3], IN={1, 64, 160, 200}, OCN=128, S=[2 x 2], P=[1 x 1], BIAS, OCV/CPU)|10.373|9.879|1.05|
|conv::Conv::(GFLOPS=1.182, K=[3 x 3], IN={1, 32, 320, 400}, OCN=64, S=[2 x 2], P=[1 x 1], BIAS, OCV/CPU)|12.782|11.624|1.10|
|conv::Conv::(GFLOPS=1.195, K=[9 x 9], IN={1, 32, 240, 320}, OCN=3, P=[4 x 4], BIAS, OCV/CPU)|90.931|90.552|1.00|
|conv::Conv::(GFLOPS=1.196, K=[3 x 3], IN={1, 384, 26, 26}, OCN=256, P=[1 x 1], OCV/CPU)|6.091|5.818|1.05|
|conv::Conv::(GFLOPS=1.210, K=[3 x 3], IN={1, 32, 256, 256}, OCN=32, PM=SAME, OCV/CPU)|7.083|6.643|1.07|
|conv::Conv::(GFLOPS=1.245, K=[3 x 3], IN={1, 64, 75, 75}, OCN=192, PM=SAME, BIAS, OCV/CPU)|5.054|5.059|1.00|
|conv::Conv::(GFLOPS=1.245, K=[3 x 3], IN={1, 96, 75, 100}, OCN=96, PM=SAME, BIAS, OCV/CPU)|5.005|4.931|1.02|
|conv::Conv::(GFLOPS=1.248, K=[3 x 3], IN={1, 256, 46, 46}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|4.951|5.065|0.98|
|conv::Conv::(GFLOPS=1.258, K=[3 x 3], IN={1, 1280, 10, 10}, OCN=546, PM=SAME, BIAS, OCV/CPU)|11.957|11.293|1.06|
|conv::Conv::(GFLOPS=1.261, K=[3 x 3], IN={1, 192, 38, 50}, OCN=192, PM=SAME, BIAS, OCV/CPU)|5.328|5.250|1.01|
|conv::Conv::(GFLOPS=1.416, K=[3 x 3], IN={1, 128, 62, 82}, OCN=128, BIAS, OCV/CPU)|5.544|5.292|1.05|
|conv::Conv::(GFLOPS=1.500, K=[3 x 3], IN={1, 128, 64, 84}, OCN=128, BIAS, OCV/CPU)|6.186|5.893|1.05|
|conv::Conv::(GFLOPS=1.586, K=[3 x 3], IN={1, 128, 66, 86}, OCN=128, BIAS, OCV/CPU)|6.153|5.834|1.05|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 256, 26, 26}, OCN=512, P=[1 x 1], OCV/CPU)|8.154|8.107|1.01|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 256, 52, 52}, OCN=512, S=[2 x 2], P=[1 x 1], OCV/CPU)|12.699|12.256|1.04|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 512, 13, 13}, OCN=1024, P=[1 x 1], OCV/CPU)|11.355|11.217|1.01|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 512, 26, 26}, OCN=1024, S=[2 x 2], P=[1 x 1], OCV/CPU)|19.062|17.814|1.07|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 64, 104, 104}, OCN=128, P=[1 x 1], OCV/CPU)|6.820|6.531|1.04|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 64, 208, 208}, OCN=128, S=[2 x 2], P=[1 x 1], OCV/CPU)|14.502|13.483|1.08|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 128, 52, 52}, OCN=256, P=[1 x 1], OCV/CPU)|6.270|6.123|1.02|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 128, 104, 104}, OCN=256, S=[2 x 2], P=[1 x 1], OCV/CPU)|13.173|12.451|1.06|
|conv::Conv::(GFLOPS=1.598, K=[3 x 3], IN={1, 32, 208, 208}, OCN=64, P=[1 x 1], OCV/CPU)|8.326|7.652|1.09|
|conv::Conv::(GFLOPS=1.598, K=[3 x 3], IN={1, 32, 416, 416}, OCN=64, S=[2 x 2], P=[1 x 1], OCV/CPU)|17.605|16.465|1.07|
|conv::Conv::(GFLOPS=1.659, K=[3 x 3], IN={1, 960, 10, 10}, OCN=960, PM=SAME, OCV/CPU)|15.675|14.771|1.06|
|conv::Conv::(GFLOPS=1.660, K=[3 x 3], IN={1, 128, 75, 75}, OCN=128, G=128, P=[1 x 1], BIAS, OCV/CPU)|0.420|0.423|0.99|
|conv::Conv::(GFLOPS=1.660, K=[3 x 3], IN={1, 128, 75, 75}, OCN=128, PM=SAME, OCV/CPU)|6.788|6.491|1.05|
|conv::Conv::(GFLOPS=1.675, K=[3 x 3], IN={1, 128, 68, 88}, OCN=128, BIAS, OCV/CPU)|6.456|6.168|1.05|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 256, 38, 38}, OCN=256, G=256, P=[1 x 1], BIAS, OCV/CPU)|0.263|0.261|1.01|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 256, 38, 38}, OCN=256, PM=SAME, OCV/CPU)|7.690|7.398|1.04|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, G=512, P=[1 x 1], BIAS, OCV/CPU)|0.200|0.202|0.99|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|10.542|10.464|1.01|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, PM=SAME, OCV/CPU)|10.876|10.728|1.01|
|conv::Conv::(GFLOPS=1.766, K=[3 x 3], IN={1, 128, 70, 90}, OCN=128, BIAS, OCV/CPU)|7.194|6.768|1.06|
|conv::Conv::(GFLOPS=1.859, K=[3 x 3], IN={1, 128, 72, 92}, OCN=128, BIAS, OCV/CPU)|7.099|6.731|1.05|
|conv::Conv::(GFLOPS=1.888, K=[3 x 3], IN={1, 1024, 10, 10}, OCN=1024, G=1024, P=[1 x 1], BIAS, OCV/CPU)|0.147|0.162|0.91|
|conv::Conv::(GFLOPS=1.888, K=[3 x 3], IN={1, 1024, 10, 10}, OCN=1024, PM=SAME, OCV/CPU)|18.558|17.141|1.08|
|conv::Conv::(GFLOPS=1.954, K=[3 x 3], IN={1, 128, 74, 94}, OCN=128, BIAS, OCV/CPU)|7.641|7.219|1.06|
|conv::Conv::(GFLOPS=1.995, K=[9 x 9], IN={1, 3, 320, 400}, OCN=32, P=[4 x 4], BIAS, OCV/CPU)|22.666|20.999|1.08|
|conv::Conv::(GFLOPS=2.052, K=[3 x 3], IN={1, 128, 76, 96}, OCN=128, BIAS, OCV/CPU)|8.523|7.921|1.08|
|conv::Conv::(GFLOPS=2.100, K=[3 x 3], IN={1, 144, 75, 75}, OCN=144, PM=SAME, OCV/CPU)|8.514|8.109|1.05|
|conv::Conv::(GFLOPS=2.153, K=[3 x 3], IN={1, 128, 78, 98}, OCN=128, BIAS, OCV/CPU)|8.300|7.878|1.05|
|conv::Conv::(GFLOPS=2.156, K=[3 x 3], IN={1, 576, 19, 19}, OCN=576, PM=SAME, OCV/CPU)|13.403|13.131|1.02|
|conv::Conv::(GFLOPS=2.255, K=[3 x 3], IN={1, 128, 80, 100}, OCN=128, BIAS, OCV/CPU)|8.920|8.357|1.07|
|conv::Conv::(GFLOPS=2.719, K=[3 x 3], IN={1, 96, 256, 256}, OCN=96, S=[2 x 2], PM=SAME, OCV/CPU)|28.827|27.616|1.04|
|conv::Conv::(GFLOPS=3.319, K=[3 x 3], IN={1, 128, 75, 75}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|12.895|12.670|1.02|
|conv::Conv::(GFLOPS=3.321, K=[3 x 3], IN={1, 64, 150, 150}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|14.120|13.078|1.08|
|conv::Conv::(GFLOPS=3.398, K=[7 x 7], IN={1, 128, 46, 46}, OCN=128, P=[3 x 3], BIAS, OCV/CPU)|27.541|27.582|1.00|
|conv::Conv::(GFLOPS=3.407, K=[3 x 3], IN={1, 512, 19, 19}, OCN=1024, D=[6 x 6], P=[6 x 6], BIAS, OCV/CPU)|32.367|31.140|1.04|
|conv::Conv::(GFLOPS=3.408, K=[3 x 3], IN={1, 256, 38, 38}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|14.934|14.910|1.00|
|conv::Conv::(GFLOPS=4.247, K=[3 x 3], IN={1, 480, 32, 32}, OCN=480, PM=SAME, OCV/CPU)|18.289|18.491|0.99|
|conv::Conv::(GFLOPS=4.247, K=[5 x 5], IN={1, 144, 128, 128}, OCN=144, S=[2 x 2], PM=SAME, OCV/CPU)|37.857|36.845|1.03|
|conv::Conv::(GFLOPS=4.566, K=[7 x 7], IN={1, 172, 46, 46}, OCN=128, P=[3 x 3], BIAS, OCV/CPU)|37.402|36.566|1.02|
|conv::Conv::(GFLOPS=4.993, K=[3 x 3], IN={1, 256, 46, 46}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|19.031|19.164|0.99|
|conv::Conv::(GFLOPS=4.993, K=[3 x 3], IN={1, 512, 46, 46}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|19.019|19.135|0.99|
|conv::Conv::(GFLOPS=4.994, K=[3 x 3], IN={1, 128, 92, 92}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|20.077|19.400|1.03|
|conv::Conv::(GFLOPS=4.997, K=[3 x 3], IN={1, 64, 184, 184}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|21.883|21.302|1.03|
|conv::Conv::(GFLOPS=5.780, K=[5 x 5], IN={1, 672, 32, 32}, OCN=672, S=[2 x 2], PM=SAME, OCV/CPU)|51.288|49.851|1.03|
|conv::Conv::(GFLOPS=6.116, K=[3 x 3], IN={1, 1152, 16, 16}, OCN=1152, PM=SAME, OCV/CPU)|27.349|28.359|0.96|
|conv::Conv::(GFLOPS=6.118, K=[3 x 3], IN={1, 144, 128, 128}, OCN=144, PM=SAME, OCV/CPU)|24.915|25.130|0.99|
|conv::Conv::(GFLOPS=6.637, K=[3 x 3], IN={1, 256, 75, 75}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|25.488|25.899|0.98|
|conv::Conv::(GFLOPS=6.638, K=[3 x 3], IN={1, 128, 150, 150}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|27.346|27.390|1.00|
|conv::Conv::(GFLOPS=6.641, K=[3 x 3], IN={1, 64, 150, 200}, OCN=192, PM=SAME, BIAS, OCV/CPU)|28.033|28.301|0.99|
|conv::Conv::(GFLOPS=6.641, K=[3 x 3], IN={1, 64, 300, 300}, OCN=64, P=[1 x 1], BIAS, OCV/CPU)|50.216|49.970|1.00|
|conv::Conv::(GFLOPS=6.814, K=[3 x 3], IN={1, 512, 38, 38}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|29.670|29.513|1.01|
|conv::Conv::(GFLOPS=8.025, K=[3 x 3], IN={1, 1024, 19, 19}, OCN=1206, P=[1 x 1], BIAS, OCV/CPU)|50.565|49.634|1.02|
|conv::Conv::(GFLOPS=9.986, K=[3 x 3], IN={1, 512, 46, 46}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|37.900|37.814|1.00|
|conv::Conv::(GFLOPS=9.987, K=[3 x 3], IN={1, 256, 92, 92}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|41.367|39.742|1.04|
|conv::Conv::(GFLOPS=9.989, K=[3 x 3], IN={1, 128, 184, 184}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|49.128|50.350|0.98|
|conv::Conv::(GFLOPS=9.993, K=[3 x 3], IN={1, 64, 368, 368}, OCN=64, P=[1 x 1], BIAS, OCV/CPU)|79.643|80.645|0.99|
|conv::Conv::(GFLOPS=10.087, K=[3 x 3], IN={1, 576, 38, 50}, OCN=512, PM=SAME, BIAS, OCV/CPU)|41.439|40.895|1.01|
|conv::Conv::(GFLOPS=10.701, K=[3 x 3], IN={1, 512, 38, 38}, OCN=804, P=[1 x 1], BIAS, OCV/CPU)|46.504|46.220|1.01|
|conv::Conv::(GFLOPS=11.797, K=[5 x 5], IN={1, 240, 64, 64}, OCN=240, PM=SAME, OCV/CPU)|98.086|96.842|1.01|
|conv::Conv::(GFLOPS=11.797, K=[5 x 5], IN={1, 480, 32, 32}, OCN=480, PM=SAME, OCV/CPU)|102.447|97.299|1.05|
|conv::Conv::(GFLOPS=16.987, K=[5 x 5], IN={1, 1152, 16, 16}, OCN=1152, PM=SAME, OCV/CPU)|145.047|144.996|1.00|
|conv::Conv::(GFLOPS=23.122, K=[5 x 5], IN={1, 672, 32, 32}, OCN=672, PM=SAME, OCV/CPU)|206.104|195.543|1.05|
### Test on M1(ARM) platform
|Name of Test|4.x|patch|4.x vs patch (x-factor)|
|---|:-:|:-:|:-:|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 2, 19}, OCN=2, G=2, S=2, P=(1, 1), BIAS, OCV/CPU)|0.001|0.001|0.97|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 2, 25}, OCN=2, G=2, P=(2, 2), PM=SAME, OCV/CPU)|0.001|0.001|0.94|
|conv1d::Conv1D::(GFLOPS=0.000, K=[3], IN={1, 6, 10}, OCN=6, PM=VALID, BIAS, OCV/CPU)|0.002|0.002|0.92|
|conv3d::Conv3D::(GFLOPS=0.000, K=[1 x 1 x 1], IN={1, 4, 9, 10, 10}, OCN=4, S=[1 x 1 x 2], P=(1, 1) x (1, 1) x (1, 1), PM=VALID, OCV/CPU)|0.003|0.003|1.00|
|conv3d::Conv3D::(GFLOPS=0.000, K=[1 x 1 x 1], IN={1, 8, 1, 10, 10}, OCN=8, G=8, P=(1, 1) x (1, 1) x (1, 1), BIAS, OCV/CPU)|0.003|0.003|1.00|
|conv3d::Conv3D::(GFLOPS=0.000, K=[3 x 3 x 3], IN={1, 2, 19, 19, 19}, OCN=2, G=2, S=[2 x 2 x 2], P=(1, 1) x (1, 1) x (1, 1), BIAS, OCV/CPU)|0.031|0.031|1.00|
|conv3d::Conv3D::(GFLOPS=0.000, K=[3 x 4 x 2], IN={1, 4, 8, 10, 10}, OCN=4, G=4, S=[1 x 2 x 1], BIAS, OCV/CPU)|0.009|0.009|1.00|
|conv3d::Conv3D::(GFLOPS=0.001, K=[3 x 3 x 3], IN={1, 2, 25, 19, 19}, OCN=2, G=2, S=[1 x 2 x 2], P=(2, 2) x (2, 2) x (2, 2), PM=SAME, OCV/CPU)|0.066|0.066|1.01|
|conv3d::Conv3D::(GFLOPS=0.002, K=[3 x 1 x 4], IN={1, 14, 5, 10, 10}, OCN=14, PM=SAME, OCV/CPU)|0.102|0.102|1.00|
|conv3d::Conv3D::(GFLOPS=0.006, K=[5 x 5 x 5], IN={1, 4, 50, 19, 19}, OCN=4, S=[2 x 2 x 2], P=(1, 1) x (1, 1) x (1, 1), PM=VALID, OCV/CPU)|0.328|0.328|1.00|
|conv3d::Conv3D::(GFLOPS=0.027, K=[3 x 3 x 3], IN={1, 6, 10, 38, 50}, OCN=6, PM=VALID, BIAS, OCV/CPU)|0.693|0.747|0.93|
|conv3d::Conv3D::(GFLOPS=0.030, K=[5 x 5 x 5], IN={1, 6, 19, 19, 19}, OCN=6, G=2, OCV/CPU)|1.268|1.266|1.00|
|conv3d::Conv3D::(GFLOPS=0.045, K=[7 x 7 x 7], IN={1, 2, 38, 38, 38}, OCN=2, S=[1 x 2 x 1], OCV/CPU)|3.530|3.581|0.99|
|conv3d::Conv3D::(GFLOPS=0.053, K=[3 x 3 x 3], IN={1, 10, 98, 10, 10}, OCN=10, PM=SAME, OCV/CPU)|1.186|1.188|1.00|
|conv3d::Conv3D::(GFLOPS=0.071, K=[7 x 7 x 7], IN={1, 6, 15, 19, 19}, OCN=6, S=[2 x 1 x 1], P=(3, 3) x (3, 3) x (3, 3), PM=SAME, BIAS, OCV/CPU)|2.682|2.683|1.00|
|conv3d::Conv3D::(GFLOPS=0.093, K=[5 x 5 x 5], IN={1, 4, 40, 75, 75}, OCN=4, S=[2 x 2 x 2], OCV/CPU)|4.490|4.501|1.00|
|conv3d::Conv3D::(GFLOPS=0.116, K=[5 x 5 x 5], IN={1, 2, 21, 75, 100}, OCN=2, BIAS, OCV/CPU)|8.914|8.938|1.00|
|conv3d::Conv3D::(GFLOPS=1.267, K=[5 x 5 x 5], IN={1, 3, 75, 75, 100}, OCN=3, PM=SAME, BIAS, OCV/CPU)|69.819|69.876|1.00|
|conv3d::Conv3D::(GFLOPS=1.343, K=[3 x 3 x 3], IN={1, 11, 9, 150, 200}, OCN=11, PM=VALID, BIAS, OCV/CPU)|24.058|22.420|1.07|
|conv::Conv::(GFLOPS=0.177, K=[1 x 1], IN={1, 512, 26, 26}, OCN=256, OCV/CPU)|2.240|2.236|1.00|
|conv::Conv::(GFLOPS=0.177, K=[1 x 1], IN={1, 1024, 13, 13}, OCN=512, OCV/CPU)|3.132|3.136|1.00|
|conv::Conv::(GFLOPS=0.178, K=[1 x 1], IN={1, 256, 52, 52}, OCN=128, OCV/CPU)|1.920|1.919|1.00|
|conv::Conv::(GFLOPS=0.210, K=[1 x 1], IN={1, 576, 38, 50}, OCN=96, PM=SAME, BIAS, OCV/CPU)|2.343|2.346|1.00|
|conv::Conv::(GFLOPS=0.231, K=[3 x 3], IN={1, 128, 56, 56}, OCN=32, P=[1 x 1], OCV/CPU)|1.234|1.116|1.11|
|conv::Conv::(GFLOPS=0.231, K=[3 x 3], IN={1, 256, 14, 14}, OCN=256, P=[1 x 1], OCV/CPU)|1.109|1.121|0.99|
|conv::Conv::(GFLOPS=0.280, K=[1 x 1], IN={1, 576, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|3.197|3.084|1.04|
|conv::Conv::(GFLOPS=0.302, K=[3 x 3], IN={1, 64, 64, 64}, OCN=64, PM=SAME, OCV/CPU)|1.123|1.148|0.98|
|conv::Conv::(GFLOPS=0.357, K=[1 x 1], IN={1, 64, 208, 208}, OCN=64, OCV/CPU)|4.836|5.061|0.96|
|conv::Conv::(GFLOPS=0.420, K=[3 x 3], IN={1, 96, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|1.535|1.463|1.05|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 128, 40, 40}, OCN=128, PM=SAME, OCV/CPU)|1.756|1.584|1.11|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 256, 20, 20}, OCN=256, PM=SAME, OCV/CPU)|1.821|1.820|1.00|
|conv::Conv::(GFLOPS=0.472, K=[3 x 3], IN={1, 512, 10, 10}, OCN=512, PM=SAME, OCV/CPU)|7.049|6.672|1.06|
|conv::Conv::(GFLOPS=0.561, K=[3 x 3], IN={1, 128, 38, 50}, OCN=128, PM=SAME, BIAS, OCV/CPU)|1.967|1.922|1.02|
|conv::Conv::(GFLOPS=0.624, K=[3 x 3], IN={1, 128, 46, 46}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|1.943|1.977|0.98|
|conv::Conv::(GFLOPS=0.701, K=[3 x 3], IN={1, 128, 38, 50}, OCN=160, PM=SAME, BIAS, OCV/CPU)|2.464|2.310|1.07|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 64, 104, 104}, OCN=64, P=[1 x 1], OCV/CPU)|2.860|2.904|0.98|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 128, 52, 52}, OCN=128, P=[1 x 1], OCV/CPU)|2.428|2.483|0.98|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 256, 26, 26}, OCN=256, P=[1 x 1], OCV/CPU)|2.955|2.983|0.99|
|conv::Conv::(GFLOPS=0.798, K=[3 x 3], IN={1, 512, 13, 13}, OCN=512, P=[1 x 1], OCV/CPU)|4.328|4.484|0.97|
|conv::Conv::(GFLOPS=0.830, K=[3 x 3], IN={1, 64, 75, 100}, OCN=96, PM=SAME, BIAS, OCV/CPU)|2.712|2.778|0.98|
|conv::Conv::(GFLOPS=0.958, K=[3 x 3], IN={1, 192, 38, 38}, OCN=192, PM=SAME, OCV/CPU)|3.205|3.331|0.96|
|conv::Conv::(GFLOPS=0.958, K=[3 x 3], IN={1, 384, 19, 19}, OCN=384, PM=SAME, OCV/CPU)|4.193|4.412|0.95|
|conv::Conv::(GFLOPS=1.022, K=[3 x 3], IN={1, 576, 19, 19}, OCN=273, PM=SAME, BIAS, OCV/CPU)|5.026|4.565|1.10|
|conv::Conv::(GFLOPS=1.112, K=[3 x 3], IN={1, 512, 10, 10}, OCN=1206, P=[1 x 1], BIAS, OCV/CPU)|14.490|14.213|1.02|
|conv::Conv::(GFLOPS=1.181, K=[3 x 3], IN={1, 64, 160, 200}, OCN=128, S=[2 x 2], P=[1 x 1], BIAS, OCV/CPU)|14.886|14.003|1.06|
|conv::Conv::(GFLOPS=1.182, K=[3 x 3], IN={1, 32, 320, 400}, OCN=64, S=[2 x 2], P=[1 x 1], BIAS, OCV/CPU)|15.923|15.184|1.05|
|conv::Conv::(GFLOPS=1.195, K=[9 x 9], IN={1, 32, 240, 320}, OCN=3, P=[4 x 4], BIAS, OCV/CPU)|45.136|41.696|1.08|
|conv::Conv::(GFLOPS=1.196, K=[3 x 3], IN={1, 384, 26, 26}, OCN=256, P=[1 x 1], OCV/CPU)|4.995|4.631|1.08|
|conv::Conv::(GFLOPS=1.210, K=[3 x 3], IN={1, 32, 256, 256}, OCN=32, PM=SAME, OCV/CPU)|6.402|6.261|1.02|
|conv::Conv::(GFLOPS=1.245, K=[3 x 3], IN={1, 64, 75, 75}, OCN=192, PM=SAME, BIAS, OCV/CPU)|4.478|3.965|1.13|
|conv::Conv::(GFLOPS=1.245, K=[3 x 3], IN={1, 96, 75, 100}, OCN=96, PM=SAME, BIAS, OCV/CPU)|3.908|3.978|0.98|
|conv::Conv::(GFLOPS=1.248, K=[3 x 3], IN={1, 256, 46, 46}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|4.176|4.206|0.99|
|conv::Conv::(GFLOPS=1.258, K=[3 x 3], IN={1, 1280, 10, 10}, OCN=546, PM=SAME, BIAS, OCV/CPU)|21.509|21.136|1.02|
|conv::Conv::(GFLOPS=1.261, K=[3 x 3], IN={1, 192, 38, 50}, OCN=192, PM=SAME, BIAS, OCV/CPU)|4.426|4.082|1.08|
|conv::Conv::(GFLOPS=1.416, K=[3 x 3], IN={1, 128, 62, 82}, OCN=128, BIAS, OCV/CPU)|4.098|4.289|0.96|
|conv::Conv::(GFLOPS=1.500, K=[3 x 3], IN={1, 128, 64, 84}, OCN=128, BIAS, OCV/CPU)|4.646|5.105|0.91|
|conv::Conv::(GFLOPS=1.586, K=[3 x 3], IN={1, 128, 66, 86}, OCN=128, BIAS, OCV/CPU)|4.746|4.724|1.00|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 256, 26, 26}, OCN=512, P=[1 x 1], OCV/CPU)|5.614|5.779|0.97|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 256, 52, 52}, OCN=512, S=[2 x 2], P=[1 x 1], OCV/CPU)|21.909|20.718|1.06|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 512, 13, 13}, OCN=1024, P=[1 x 1], OCV/CPU)|8.256|8.290|1.00|
|conv::Conv::(GFLOPS=1.595, K=[3 x 3], IN={1, 512, 26, 26}, OCN=1024, S=[2 x 2], P=[1 x 1], OCV/CPU)|25.196|23.267|1.08|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 64, 104, 104}, OCN=128, P=[1 x 1], OCV/CPU)|5.721|5.172|1.11|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 64, 208, 208}, OCN=128, S=[2 x 2], P=[1 x 1], OCV/CPU)|20.066|18.322|1.10|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 128, 52, 52}, OCN=256, P=[1 x 1], OCV/CPU)|4.448|4.542|0.98|
|conv::Conv::(GFLOPS=1.596, K=[3 x 3], IN={1, 128, 104, 104}, OCN=256, S=[2 x 2], P=[1 x 1], OCV/CPU)|19.193|19.013|1.01|
|conv::Conv::(GFLOPS=1.598, K=[3 x 3], IN={1, 32, 208, 208}, OCN=64, P=[1 x 1], OCV/CPU)|6.009|5.964|1.01|
|conv::Conv::(GFLOPS=1.598, K=[3 x 3], IN={1, 32, 416, 416}, OCN=64, S=[2 x 2], P=[1 x 1], OCV/CPU)|20.169|20.009|1.01|
|conv::Conv::(GFLOPS=1.659, K=[3 x 3], IN={1, 960, 10, 10}, OCN=960, PM=SAME, OCV/CPU)|22.584|23.423|0.96|
|conv::Conv::(GFLOPS=1.660, K=[3 x 3], IN={1, 128, 75, 75}, OCN=128, G=128, P=[1 x 1], BIAS, OCV/CPU)|0.372|0.504|0.74|
|conv::Conv::(GFLOPS=1.660, K=[3 x 3], IN={1, 128, 75, 75}, OCN=128, PM=SAME, OCV/CPU)|5.426|5.456|0.99|
|conv::Conv::(GFLOPS=1.675, K=[3 x 3], IN={1, 128, 68, 88}, OCN=128, BIAS, OCV/CPU)|4.945|5.221|0.95|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 256, 38, 38}, OCN=256, G=256, P=[1 x 1], BIAS, OCV/CPU)|0.210|0.261|0.81|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 256, 38, 38}, OCN=256, PM=SAME, OCV/CPU)|5.720|5.997|0.95|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, G=512, P=[1 x 1], BIAS, OCV/CPU)|0.149|0.161|0.93|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|7.154|7.225|0.99|
|conv::Conv::(GFLOPS=1.704, K=[3 x 3], IN={1, 512, 19, 19}, OCN=512, PM=SAME, OCV/CPU)|7.184|7.223|0.99|
|conv::Conv::(GFLOPS=1.766, K=[3 x 3], IN={1, 128, 70, 90}, OCN=128, BIAS, OCV/CPU)|5.324|5.343|1.00|
|conv::Conv::(GFLOPS=1.859, K=[3 x 3], IN={1, 128, 72, 92}, OCN=128, BIAS, OCV/CPU)|5.114|5.238|0.98|
|conv::Conv::(GFLOPS=1.888, K=[3 x 3], IN={1, 1024, 10, 10}, OCN=1024, G=1024, P=[1 x 1], BIAS, OCV/CPU)|0.111|0.121|0.92|
|conv::Conv::(GFLOPS=1.888, K=[3 x 3], IN={1, 1024, 10, 10}, OCN=1024, PM=SAME, OCV/CPU)|25.907|26.804|0.97|
|conv::Conv::(GFLOPS=1.954, K=[3 x 3], IN={1, 128, 74, 94}, OCN=128, BIAS, OCV/CPU)|5.695|5.654|1.01|
|conv::Conv::(GFLOPS=1.995, K=[9 x 9], IN={1, 3, 320, 400}, OCN=32, P=[4 x 4], BIAS, OCV/CPU)|27.435|27.566|1.00|
|conv::Conv::(GFLOPS=2.052, K=[3 x 3], IN={1, 128, 76, 96}, OCN=128, BIAS, OCV/CPU)|6.944|6.164|1.13|
|conv::Conv::(GFLOPS=2.100, K=[3 x 3], IN={1, 144, 75, 75}, OCN=144, PM=SAME, OCV/CPU)|7.180|6.717|1.07|
|conv::Conv::(GFLOPS=2.153, K=[3 x 3], IN={1, 128, 78, 98}, OCN=128, BIAS, OCV/CPU)|6.817|6.050|1.13|
|conv::Conv::(GFLOPS=2.156, K=[3 x 3], IN={1, 576, 19, 19}, OCN=576, PM=SAME, OCV/CPU)|9.225|8.660|1.07|
|conv::Conv::(GFLOPS=2.255, K=[3 x 3], IN={1, 128, 80, 100}, OCN=128, BIAS, OCV/CPU)|7.496|6.625|1.13|
|conv::Conv::(GFLOPS=2.719, K=[3 x 3], IN={1, 96, 256, 256}, OCN=96, S=[2 x 2], PM=SAME, OCV/CPU)|35.520|36.056|0.99|
|conv::Conv::(GFLOPS=3.319, K=[3 x 3], IN={1, 128, 75, 75}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|9.990|9.702|1.03|
|conv::Conv::(GFLOPS=3.321, K=[3 x 3], IN={1, 64, 150, 150}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|10.517|10.746|0.98|
|conv::Conv::(GFLOPS=3.398, K=[7 x 7], IN={1, 128, 46, 46}, OCN=128, P=[3 x 3], BIAS, OCV/CPU)|36.702|36.731|1.00|
|conv::Conv::(GFLOPS=3.407, K=[3 x 3], IN={1, 512, 19, 19}, OCN=1024, D=[6 x 6], P=[6 x 6], BIAS, OCV/CPU)|41.035|38.280|1.07|
|conv::Conv::(GFLOPS=3.408, K=[3 x 3], IN={1, 256, 38, 38}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|10.981|10.573|1.04|
|conv::Conv::(GFLOPS=4.247, K=[3 x 3], IN={1, 480, 32, 32}, OCN=480, PM=SAME, OCV/CPU)|12.863|12.384|1.04|
|conv::Conv::(GFLOPS=4.247, K=[5 x 5], IN={1, 144, 128, 128}, OCN=144, S=[2 x 2], PM=SAME, OCV/CPU)|50.437|54.088|0.93|
|conv::Conv::(GFLOPS=4.566, K=[7 x 7], IN={1, 172, 46, 46}, OCN=128, P=[3 x 3], BIAS, OCV/CPU)|50.650|50.635|1.00|
|conv::Conv::(GFLOPS=4.993, K=[3 x 3], IN={1, 256, 46, 46}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|14.696|14.606|1.01|
|conv::Conv::(GFLOPS=4.993, K=[3 x 3], IN={1, 512, 46, 46}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|16.201|15.426|1.05|
|conv::Conv::(GFLOPS=4.994, K=[3 x 3], IN={1, 128, 92, 92}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|16.061|14.292|1.12|
|conv::Conv::(GFLOPS=4.997, K=[3 x 3], IN={1, 64, 184, 184}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|17.743|18.250|0.97|
|conv::Conv::(GFLOPS=5.780, K=[5 x 5], IN={1, 672, 32, 32}, OCN=672, S=[2 x 2], PM=SAME, OCV/CPU)|77.909|78.165|1.00|
|conv::Conv::(GFLOPS=6.116, K=[3 x 3], IN={1, 1152, 16, 16}, OCN=1152, PM=SAME, OCV/CPU)|21.579|21.879|0.99|
|conv::Conv::(GFLOPS=6.118, K=[3 x 3], IN={1, 144, 128, 128}, OCN=144, PM=SAME, OCV/CPU)|20.424|19.589|1.04|
|conv::Conv::(GFLOPS=6.637, K=[3 x 3], IN={1, 256, 75, 75}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|19.389|19.461|1.00|
|conv::Conv::(GFLOPS=6.638, K=[3 x 3], IN={1, 128, 150, 150}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|21.319|20.358|1.05|
|conv::Conv::(GFLOPS=6.641, K=[3 x 3], IN={1, 64, 150, 200}, OCN=192, PM=SAME, BIAS, OCV/CPU)|22.609|21.826|1.04|
|conv::Conv::(GFLOPS=6.641, K=[3 x 3], IN={1, 64, 300, 300}, OCN=64, P=[1 x 1], BIAS, OCV/CPU)|25.497|25.789|0.99|
|conv::Conv::(GFLOPS=6.814, K=[3 x 3], IN={1, 512, 38, 38}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|21.966|22.108|0.99|
|conv::Conv::(GFLOPS=8.025, K=[3 x 3], IN={1, 1024, 19, 19}, OCN=1206, P=[1 x 1], BIAS, OCV/CPU)|35.883|33.470|1.07|
|conv::Conv::(GFLOPS=9.986, K=[3 x 3], IN={1, 512, 46, 46}, OCN=512, P=[1 x 1], BIAS, OCV/CPU)|31.041|29.314|1.06|
|conv::Conv::(GFLOPS=9.987, K=[3 x 3], IN={1, 256, 92, 92}, OCN=256, P=[1 x 1], BIAS, OCV/CPU)|29.922|28.145|1.06|
|conv::Conv::(GFLOPS=9.989, K=[3 x 3], IN={1, 128, 184, 184}, OCN=128, P=[1 x 1], BIAS, OCV/CPU)|31.624|31.148|1.02|
|conv::Conv::(GFLOPS=9.993, K=[3 x 3], IN={1, 64, 368, 368}, OCN=64, P=[1 x 1], BIAS, OCV/CPU)|38.564|39.164|0.98|
|conv::Conv::(GFLOPS=10.087, K=[3 x 3], IN={1, 576, 38, 50}, OCN=512, PM=SAME, BIAS, OCV/CPU)|31.502|30.269|1.04|
|conv::Conv::(GFLOPS=10.701, K=[3 x 3], IN={1, 512, 38, 38}, OCN=804, P=[1 x 1], BIAS, OCV/CPU)|34.248|34.589|0.99|
|conv::Conv::(GFLOPS=11.797, K=[5 x 5], IN={1, 240, 64, 64}, OCN=240, PM=SAME, OCV/CPU)|130.211|134.120|0.97|
|conv::Conv::(GFLOPS=11.797, K=[5 x 5], IN={1, 480, 32, 32}, OCN=480, PM=SAME, OCV/CPU)|127.490|132.874|0.96|
|conv::Conv::(GFLOPS=16.987, K=[5 x 5], IN={1, 1152, 16, 16}, OCN=1152, PM=SAME, OCV/CPU)|199.834|200.081|1.00|
|conv::Conv::(GFLOPS=23.122, K=[5 x 5], IN={1, 672, 32, 32}, OCN=672, PM=SAME, OCV/CPU)|247.346|247.523|1.00|
### Pull Request Readiness Checklist
See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request
- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [ ] There is a reference to the original bug report and related work
- [ ] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
Patch to opencv_extra has the same branch name.
- [ ] The feature is well documented and sample code can be built with the project CMake
```
force_builders=Linux AVX2,Custom Win
build_image:Custom Win=msvs2019
CPU_BASELINE:Custom Win=AVX512_SKX
```
2 years ago
ocv_add_dispatched_file_force_all ( "layers/cpu_kernels/conv_depthwise" AVX AVX2 RVV LASX )
ocv_add_dispatched_file_force_all ( "layers/cpu_kernels/conv_winograd_f63" AVX AVX2 NEON_FP16 )
ocv_add_dispatched_file_force_all ( "layers/cpu_kernels/fast_gemm_kernels" AVX AVX2 NEON LASX )
ocv_add_module ( dnn opencv_core opencv_imgproc WRAP python java objc js )
include ( ${ CMAKE_CURRENT_LIST_DIR } /cmake/plugin.cmake )
ocv_option ( OPENCV_DNN_OPENCL "Build with OpenCL support" HAVE_OPENCL AND NOT APPLE )
if ( OPENCV_DNN_OPENCL AND HAVE_OPENCL )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "CV_OCL4DNN=1" )
endif ( )
if ( WITH_WEBNN AND HAVE_WEBNN )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "HAVE_WEBNN=1" )
endif ( )
if ( HAVE_TIMVX )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "HAVE_TIMVX=1" )
endif ( )
if ( HAVE_CANN )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "HAVE_CANN=1" )
endif ( )
ocv_option ( OPENCV_DNN_CUDA "Build with CUDA support"
H A V E _ C U D A
A N D H A V E _ C U B L A S
A N D H A V E _ C U D N N
)
if ( OPENCV_DNN_CUDA )
if ( HAVE_CUDA AND HAVE_CUBLAS AND HAVE_CUDNN )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "CV_CUDA4DNN=1" )
else ( )
if ( NOT HAVE_CUDA )
message ( SEND_ERROR "DNN: CUDA backend requires CUDA Toolkit. Please resolve dependency or disable OPENCV_DNN_CUDA=OFF" )
elseif ( NOT HAVE_CUBLAS )
message ( SEND_ERROR "DNN: CUDA backend requires cuBLAS. Please resolve dependency or disable OPENCV_DNN_CUDA=OFF" )
elseif ( NOT HAVE_CUDNN )
message ( SEND_ERROR "DNN: CUDA backend requires cuDNN. Please resolve dependency or disable OPENCV_DNN_CUDA=OFF" )
endif ( )
endif ( )
endif ( )
ocv_cmake_hook_append ( INIT_MODULE_SOURCES_opencv_dnn "${CMAKE_CURRENT_LIST_DIR}/cmake/hooks/INIT_MODULE_SOURCES_opencv_dnn.cmake" )
if ( MSVC )
add_definitions ( -D_CRT_SECURE_NO_WARNINGS=1 )
ocv_warnings_disable ( CMAKE_CXX_FLAGS /wd4244 /wd4267 /wd4018 /wd4355 /wd4800 /wd4251 /wd4996 /wd4146
/ w d 4 3 0 5 / w d 4 1 2 7 / w d 4 1 0 0 / w d 4 5 1 2 / w d 4 1 2 5 / w d 4 3 8 9 / w d 4 5 1 0 / w d 4 6 1 0
/ w d 4 7 0 2 / w d 4 4 5 6 / w d 4 4 5 7 / w d 4 0 6 5 / w d 4 3 1 0 / w d 4 6 6 1 / w d 4 5 0 6
)
if ( MSVC_VERSION LESS 1920 ) # MSVS 2015/2017, .pb.cc generated files
ocv_warnings_disable ( CMAKE_CXX_FLAGS /wd4309 ) # 'static_cast': truncation of constant value
endif ( )
if ( MSVC_VERSION LESS 1920 ) # <MSVS2019, .pb.cc generated files
ocv_warnings_disable ( CMAKE_CXX_FLAGS /wd4189 ) # local variable is initialized but not referenced
ocv_warnings_disable ( CMAKE_CXX_FLAGS /wd4592 ) # symbol will be dynamically initialized (implementation limitation)
endif ( )
else ( )
ocv_warnings_disable ( CMAKE_CXX_FLAGS -Wno-deprecated -Wmissing-prototypes -Wmissing-declarations -Wshadow
- W u n u s e d - p a r a m e t e r - W s i g n - c o m p a r e
)
endif ( )
if ( HAVE_CUDA )
ocv_warnings_disable ( CMAKE_CXX_FLAGS -Wundef )
endif ( )
if ( NOT HAVE_CXX11 )
ocv_warnings_disable ( CMAKE_CXX_FLAGS -Wno-undef ) # LANG_CXX11 from protobuf files
endif ( )
if ( APPLE_FRAMEWORK )
ocv_warnings_disable ( CMAKE_CXX_FLAGS -Wshorten-64-to-32 )
endif ( )
if ( ANDROID )
add_definitions ( -DDISABLE_POSIX_MEMALIGN -DTH_DISABLE_HEAP_TRACKING )
endif ( )
if ( NOT BUILD_PROTOBUF )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "OPENCV_DNN_EXTERNAL_PROTOBUF=1" )
endif ( )
#suppress warnings in autogenerated caffe.pb.* files
ocv_warnings_disable ( CMAKE_CXX_FLAGS
/ w d 4 1 2 5 / w d 4 2 6 7 / w d 4 1 2 7 / w d 4 2 4 4 / w d 4 5 1 2 / w d 4 7 0 2
/ w d 4 4 5 6 / w d 4 5 1 0 / w d 4 6 1 0 / w d 4 8 0 0
/ w d 4 7 0 1 / w d 4 7 0 3 # potentially uninitialized local/pointer variable 'value' used
/ w d 4 5 0 5 # unreferenced local function has been removed
/ w d 4 4 5 8 # declaration of 'x' hides class member. GCC still works, MSVC bug is here: https://developercommunity.visualstudio.com/content/problem/219311/c-c4458-declaration-hides-class-member-warning-iss.html
- w d 8 5 8 - w d 2 1 9 6
- W i n v a l i d - o f f s e t o f # Apple Clang (attr_value.pb.cc)
)
set ( include_dirs "" )
set ( libs "" )
if ( HAVE_PROTOBUF )
ocv_target_compile_definitions ( ${ the_module } PRIVATE "HAVE_PROTOBUF=1" )
if ( PROTOBUF_UPDATE_FILES )
file ( GLOB proto_files "${CMAKE_CURRENT_LIST_DIR}/src/tensorflow/*.proto" "${CMAKE_CURRENT_LIST_DIR}/src/caffe/opencv-caffe.proto" "${CMAKE_CURRENT_LIST_DIR}/src/onnx/opencv-onnx.proto" )
set ( PROTOBUF_GENERATE_CPP_APPEND_PATH ON ) # required for tensorflow
protobuf_generate_cpp ( fw_srcs fw_hdrs ${ proto_files } )
else ( )
file ( GLOB fw_srcs "${CMAKE_CURRENT_LIST_DIR}/misc/tensorflow/*.cc" "${CMAKE_CURRENT_LIST_DIR}/misc/caffe/opencv-caffe.pb.cc" "${CMAKE_CURRENT_LIST_DIR}/misc/onnx/opencv-onnx.pb.cc" )
file ( GLOB fw_hdrs "${CMAKE_CURRENT_LIST_DIR}/misc/tensorflow/*.h" "${CMAKE_CURRENT_LIST_DIR}/misc/caffe/opencv-caffe.pb.h" "${CMAKE_CURRENT_LIST_DIR}/misc/onnx/opencv-onnx.pb.h" )
set ( fw_inc "${CMAKE_CURRENT_LIST_DIR}/misc/caffe" "${CMAKE_CURRENT_LIST_DIR}/misc/tensorflow" "${CMAKE_CURRENT_LIST_DIR}/misc/onnx" )
endif ( )
endif ( )
ocv_option ( OPENCV_DNN_TFLITE "Build with TFLite support" ( TARGET ocv.3rdparty.flatbuffers ) )
if ( TARGET ocv.3rdparty.flatbuffers AND OPENCV_DNN_TFLITE )
if ( NOT HAVE_FLATBUFFERS )
message ( FATAL_ERROR "DNN: TFLite is not supported without enabled 'flatbuffers'. Check build configuration." )
endif ( )
list ( APPEND libs ocv.3rdparty.flatbuffers )
list ( APPEND fw_hdrs "${CMAKE_CURRENT_LIST_DIR}/misc/tflite/schema_generated.h" )
list ( APPEND fw_inc "${CMAKE_CURRENT_LIST_DIR}/misc/tflite" )
# Schema is generated by this command:
#add_custom_command(
# OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/schema_generated.h"
# COMMAND flatbuffers::flatc --cpp -o "${CMAKE_CURRENT_BINARY_DIR}" "${CMAKE_CURRENT_LIST_DIR}/src/tflite/schema.fbs")
endif ( )
list ( APPEND include_dirs ${ fw_inc } )
list ( APPEND libs ${ Protobuf_LIBRARIES } )
if ( NOT BUILD_PROTOBUF )
list ( APPEND include_dirs ${ Protobuf_INCLUDE_DIRS } )
endif ( )
set ( sources_options "" )
list ( APPEND libs ${ LAPACK_LIBRARIES } )
if ( OPENCV_DNN_OPENCL AND HAVE_OPENCL )
list ( APPEND include_dirs ${ OPENCL_INCLUDE_DIRS } )
else ( )
set ( sources_options EXCLUDE_OPENCL )
endif ( )
if ( OPENCV_DNN_CUDA AND HAVE_CUDA AND HAVE_CUBLAS AND HAVE_CUDNN )
list ( APPEND include_dirs ${ CUDA_TOOLKIT_INCLUDE } ${ CUDNN_INCLUDE_DIRS } )
Merge pull request #14827 from YashasSamaga:cuda4dnn-csl-low
CUDA backend for the DNN module
* stub cuda4dnn design
* minor fixes for tests and doxygen
* add csl public api directory to module headers
* add low-level CSL components
* add high-level CSL components
* integrate csl::Tensor into backbone code
* switch to CPU iff unsupported; otherwise, fail on error
* add fully connected layer
* add softmax layer
* add activation layers
* support arbitary rank TensorDescriptor
* pass input wrappers to `initCUDA()`
* add 1d/2d/3d-convolution
* add pooling layer
* reorganize and refactor code
* fixes for gcc, clang and doxygen; remove cxx14/17 code
* add blank_layer
* add LRN layer
* add rounding modes for pooling layer
* split tensor.hpp into tensor.hpp and tensor_ops.hpp
* add concat layer
* add scale layer
* add batch normalization layer
* split math.cu into activations.cu and math.hpp
* add eltwise layer
* add flatten layer
* add tensor transform api
* add asymmetric padding support for convolution layer
* add reshape layer
* fix rebase issues
* add permute layer
* add padding support for concat layer
* refactor and reorganize code
* add normalize layer
* optimize bias addition in scale layer
* add prior box layer
* fix and optimize normalize layer
* add asymmetric padding support for pooling layer
* add event API
* improve pooling performance for some padding scenarios
* avoid over-allocation of compute resources to kernels
* improve prior box performance
* enable layer fusion
* add const layer
* add resize layer
* add slice layer
* add padding layer
* add deconvolution layer
* fix channelwise ReLU initialization
* add vector traits
* add vectorized versions of relu, clipped_relu, power
* add vectorized concat kernels
* improve concat_with_offsets performance
* vectorize scale and bias kernels
* add support for multi-billion element tensors
* vectorize prior box kernels
* fix address alignment check
* improve bias addition performance of conv/deconv/fc layers
* restructure code for supporting multiple targets
* add DNN_TARGET_CUDA_FP64
* add DNN_TARGET_FP16
* improve vectorization
* add region layer
* improve tensor API, add dynamic ranks
1. use ManagedPtr instead of a Tensor in backend wrapper
2. add new methods to tensor classes
- size_range: computes the combined size of for a given axis range
- tensor span/view can be constructed from a raw pointer and shape
3. the tensor classes can change their rank at runtime (previously rank was fixed at compile-time)
4. remove device code from tensor classes (as they are unused)
5. enforce strict conditions on tensor class APIs to improve debugging ability
* fix parametric relu activation
* add squeeze/unsqueeze tensor API
* add reorg layer
* optimize permute and enable 2d permute
* enable 1d and 2d slice
* add split layer
* add shuffle channel layer
* allow tensors of different ranks in reshape primitive
* patch SliceOp to allow Crop Layer
* allow extra shape inputs in reshape layer
* use `std::move_backward` instead of `std::move` for insert in resizable_static_array
* improve workspace management
* add spatial LRN
* add nms (cpu) to region layer
* add max pooling with argmax ( and a fix to limits.hpp)
* add max unpooling layer
* rename DNN_TARGET_CUDA_FP32 to DNN_TARGET_CUDA
* update supportBackend to be more rigorous
* remove stray include from preventing non-cuda build
* include op_cuda.hpp outside condition #if
* refactoring, fixes and many optimizations
* drop DNN_TARGET_CUDA_FP64
* fix gcc errors
* increase max. tensor rank limit to six
* add Interp layer
* drop custom layers; use BackendNode
* vectorize activation kernels
* fixes for gcc
* remove wrong assertion
* fix broken assertion in unpooling primitive
* fix build errors in non-CUDA build
* completely remove workspace from public API
* fix permute layer
* enable accuracy and perf. tests for DNN_TARGET_CUDA
* add asynchronous forward
* vectorize eltwise ops
* vectorize fill kernel
* fixes for gcc
* remove CSL headers from public API
* remove csl header source group from cmake
* update min. cudnn version in cmake
* add numerically stable FP32 log1pexp
* refactor code
* add FP16 specialization to cudnn based tensor addition
* vectorize scale1 and bias1 + minor refactoring
* fix doxygen build
* fix invalid alignment assertion
* clear backend wrappers before allocateLayers
* ignore memory lock failures
* do not allocate internal blobs
* integrate NVTX
* add numerically stable half precision log1pexp
* fix indentation, following coding style, improve docs
* remove accidental modification of IE code
* Revert "add asynchronous forward"
This reverts commit 1154b9da9da07e9b52f8a81bdcea48cf31c56f70.
* [cmake] throw error for unsupported CC versions
* fix rebase issues
* add more docs, refactor code, fix bugs
* minor refactoring and fixes
* resolve warnings/errors from clang
* remove haveCUDA() checks from supportBackend()
* remove NVTX integration
* changes based on review comments
* avoid exception when no CUDA device is present
* add color code for CUDA in Net::dump
5 years ago
set ( CC_LIST ${ CUDA_ARCH_BIN } )
separate_arguments ( CC_LIST )
foreach ( cc ${ CC_LIST } )
if ( cc VERSION_LESS 3.0 )
message ( FATAL_ERROR "CUDA backend for DNN module requires CC 3.0 or higher. Please remove unsupported architectures from CUDA_ARCH_BIN option or disable OPENCV_DNN_CUDA=OFF." )
Merge pull request #14827 from YashasSamaga:cuda4dnn-csl-low
CUDA backend for the DNN module
* stub cuda4dnn design
* minor fixes for tests and doxygen
* add csl public api directory to module headers
* add low-level CSL components
* add high-level CSL components
* integrate csl::Tensor into backbone code
* switch to CPU iff unsupported; otherwise, fail on error
* add fully connected layer
* add softmax layer
* add activation layers
* support arbitary rank TensorDescriptor
* pass input wrappers to `initCUDA()`
* add 1d/2d/3d-convolution
* add pooling layer
* reorganize and refactor code
* fixes for gcc, clang and doxygen; remove cxx14/17 code
* add blank_layer
* add LRN layer
* add rounding modes for pooling layer
* split tensor.hpp into tensor.hpp and tensor_ops.hpp
* add concat layer
* add scale layer
* add batch normalization layer
* split math.cu into activations.cu and math.hpp
* add eltwise layer
* add flatten layer
* add tensor transform api
* add asymmetric padding support for convolution layer
* add reshape layer
* fix rebase issues
* add permute layer
* add padding support for concat layer
* refactor and reorganize code
* add normalize layer
* optimize bias addition in scale layer
* add prior box layer
* fix and optimize normalize layer
* add asymmetric padding support for pooling layer
* add event API
* improve pooling performance for some padding scenarios
* avoid over-allocation of compute resources to kernels
* improve prior box performance
* enable layer fusion
* add const layer
* add resize layer
* add slice layer
* add padding layer
* add deconvolution layer
* fix channelwise ReLU initialization
* add vector traits
* add vectorized versions of relu, clipped_relu, power
* add vectorized concat kernels
* improve concat_with_offsets performance
* vectorize scale and bias kernels
* add support for multi-billion element tensors
* vectorize prior box kernels
* fix address alignment check
* improve bias addition performance of conv/deconv/fc layers
* restructure code for supporting multiple targets
* add DNN_TARGET_CUDA_FP64
* add DNN_TARGET_FP16
* improve vectorization
* add region layer
* improve tensor API, add dynamic ranks
1. use ManagedPtr instead of a Tensor in backend wrapper
2. add new methods to tensor classes
- size_range: computes the combined size of for a given axis range
- tensor span/view can be constructed from a raw pointer and shape
3. the tensor classes can change their rank at runtime (previously rank was fixed at compile-time)
4. remove device code from tensor classes (as they are unused)
5. enforce strict conditions on tensor class APIs to improve debugging ability
* fix parametric relu activation
* add squeeze/unsqueeze tensor API
* add reorg layer
* optimize permute and enable 2d permute
* enable 1d and 2d slice
* add split layer
* add shuffle channel layer
* allow tensors of different ranks in reshape primitive
* patch SliceOp to allow Crop Layer
* allow extra shape inputs in reshape layer
* use `std::move_backward` instead of `std::move` for insert in resizable_static_array
* improve workspace management
* add spatial LRN
* add nms (cpu) to region layer
* add max pooling with argmax ( and a fix to limits.hpp)
* add max unpooling layer
* rename DNN_TARGET_CUDA_FP32 to DNN_TARGET_CUDA
* update supportBackend to be more rigorous
* remove stray include from preventing non-cuda build
* include op_cuda.hpp outside condition #if
* refactoring, fixes and many optimizations
* drop DNN_TARGET_CUDA_FP64
* fix gcc errors
* increase max. tensor rank limit to six
* add Interp layer
* drop custom layers; use BackendNode
* vectorize activation kernels
* fixes for gcc
* remove wrong assertion
* fix broken assertion in unpooling primitive
* fix build errors in non-CUDA build
* completely remove workspace from public API
* fix permute layer
* enable accuracy and perf. tests for DNN_TARGET_CUDA
* add asynchronous forward
* vectorize eltwise ops
* vectorize fill kernel
* fixes for gcc
* remove CSL headers from public API
* remove csl header source group from cmake
* update min. cudnn version in cmake
* add numerically stable FP32 log1pexp
* refactor code
* add FP16 specialization to cudnn based tensor addition
* vectorize scale1 and bias1 + minor refactoring
* fix doxygen build
* fix invalid alignment assertion
* clear backend wrappers before allocateLayers
* ignore memory lock failures
* do not allocate internal blobs
* integrate NVTX
* add numerically stable half precision log1pexp
* fix indentation, following coding style, improve docs
* remove accidental modification of IE code
* Revert "add asynchronous forward"
This reverts commit 1154b9da9da07e9b52f8a81bdcea48cf31c56f70.
* [cmake] throw error for unsupported CC versions
* fix rebase issues
* add more docs, refactor code, fix bugs
* minor refactoring and fixes
* resolve warnings/errors from clang
* remove haveCUDA() checks from supportBackend()
* remove NVTX integration
* changes based on review comments
* avoid exception when no CUDA device is present
* add color code for CUDA in Net::dump
5 years ago
endif ( )
endforeach ( )
unset ( CC_LIST )
if ( ENABLE_CUDA_FIRST_CLASS_LANGUAGE )
list ( APPEND libs ${ CUDNN_LIBRARIES } CUDA::cublas ${ CUDA_LIB_EXT } )
if ( NOT CUDA_VERSION VERSION_LESS 10.1 )
list ( APPEND libs CUDA::cublasLt ${ CUDA_LIB_EXT } )
endif ( )
endif ( )
else ( )
set ( sources_options ${ sources_options } EXCLUDE_CUDA )
endif ( )
if ( HAVE_TIMVX )
list ( APPEND include_dirs ${ TIMVX_INCLUDE_DIR } )
list ( APPEND libs -Wl,--whole-archive ${ TIMVX_LIBRARY } -Wl,--no-whole-archive )
endif ( )
if ( HAVE_CANN )
list ( APPEND include_dirs ${ CANN_INCLUDE_DIRS } )
list ( APPEND libs -Wl,--whole-archive ${ CANN_LIBRARIES } -Wl,--no-whole-archive )
endif ( )
set ( webnn_srcs "" )
if ( NOT EMSCRIPTEN )
if ( HAVE_WEBNN )
list ( APPEND include_dirs ${ WEBNN_HEADER_DIRS } )
list ( APPEND include_dirs ${ WEBNN_INCLUDE_DIRS } )
list ( APPEND libs -Wl,--whole-archive ${ WEBNN_LIBRARIES } -Wl,--no-whole-archive )
list ( APPEND webnn_srcs $ENV{ WEBNN_NATIVE_DIR } /gen/src/webnn/webnn_cpp.cpp )
endif ( )
endif ( )
ocv_module_include_directories ( ${ include_dirs } )
if ( CMAKE_CXX_COMPILER_ID STREQUAL "GNU" )
ocv_append_source_files_cxx_compiler_options ( fw_srcs "-Wno-suggest-override" ) # GCC
ocv_append_source_files_cxx_compiler_options ( fw_srcs "-Wno-array-bounds" ) # GCC 9.3.0 (Ubuntu 20.04)
elseif ( CMAKE_CXX_COMPILER_ID STREQUAL "Clang" )
ocv_append_source_files_cxx_compiler_options ( fw_srcs "-Wno-inconsistent-missing-override" ) # Clang
endif ( )
set ( dnn_runtime_libs "" )
file ( GLOB_RECURSE dnn_srcs
" $ { C M A K E _ C U R R E N T _ L I S T _ D I R } / s r c / * . c p p "
)
file ( GLOB_RECURSE dnn_int_hdrs
" $ { C M A K E _ C U R R E N T _ L I S T _ D I R } / s r c / * . h p p "
" $ { C M A K E _ C U R R E N T _ L I S T _ D I R } / s r c / * . h "
)
set ( dnn_plugin_srcs ${ dnn_srcs } ${ dnn_int_hdrs } )
ocv_list_filterout_ex ( dnn_plugin_srcs
" / s r c / d n n . c p p $ | / s r c / d n n _ u t i l s . c p p $ | / s r c / d n n _ r e a d . c p p $ | / s r c / r e g i s t r y . c p p $ | / s r c / b a c k e n d . c p p $ "
# importers
" / s r c / ( c a f f e | d a r k n e t | o n n x | t e n s o r f l o w | t o r c h ) / "
# executors
" / s r c / ( c u d a | c u d a 4 d n n | o c l 4 d n n | v k c o m | w e b n n ) / "
)
ocv_option ( OPENCV_DNN_OPENVINO "Build with OpenVINO support (2021.4+)" ( TARGET ocv.3rdparty.openvino ) )
if ( TARGET ocv.3rdparty.openvino AND OPENCV_DNN_OPENVINO )
if ( NOT HAVE_OPENVINO AND NOT HAVE_NGRAPH )
message ( FATAL_ERROR "DNN: Inference Engine is not supported without enabled 'nGraph'. Check build configuration." )
endif ( )
if ( "openvino" IN_LIST DNN_PLUGIN_LIST OR DNN_PLUGIN_LIST STREQUAL "all" )
# plugin doesn't support PCH, separate directory scope is necessary
# opencv_world requires absolute path
add_subdirectory ( "${CMAKE_CURRENT_LIST_DIR}/misc/plugin/openvino" "${CMAKE_CURRENT_BINARY_DIR}/dnn_plugin_openvino" )
elseif ( NOT OPENCV_DNN_BUILTIN_BACKEND )
list ( APPEND dnn_runtime_libs ocv.3rdparty.openvino )
endif ( )
endif ( )
set ( OPENCV_DNN_BACKEND_DEFAULT "" CACHE STRING "Default backend used by the DNN module (DNN_BACKEND_OPENCV if empty)" )
if ( OPENCV_DNN_BACKEND_DEFAULT )
ocv_append_source_file_compile_definitions ( "${CMAKE_CURRENT_LIST_DIR}/src/dnn_params.cpp" "OPENCV_DNN_BACKEND_DEFAULT=${OPENCV_DNN_BACKEND_DEFAULT}" )
endif ( )
ocv_install_used_external_targets ( ${ libs } ${ dnn_runtime_libs } )
ocv_glob_module_sources ( ${ sources_options } SOURCES ${ fw_srcs } ${ webnn_srcs } )
ocv_create_module ( ${ libs } ${ dnn_runtime_libs } )
ocv_add_samples ( )
ocv_add_accuracy_tests ( ${ dnn_runtime_libs } )
if ( NOT BUILD_PROTOBUF )
if ( TARGET opencv_test_dnn )
ocv_target_compile_definitions ( opencv_test_dnn PRIVATE "OPENCV_DNN_EXTERNAL_PROTOBUF=1" )
endif ( )
endif ( )
set ( perf_path "${CMAKE_CURRENT_LIST_DIR}/perf" )
file ( GLOB_RECURSE perf_srcs "${perf_path}/*.cpp" )
file ( GLOB_RECURSE perf_hdrs "${perf_path}/*.hpp" "${perf_path}/*.h" )
ocv_add_perf_tests ( ${ dnn_runtime_libs }
F I L E S t e s t _ c o m m o n " $ { C M A K E _ C U R R E N T _ L I S T _ D I R } / t e s t / t e s t _ c o m m o n . h p p " " $ { C M A K E _ C U R R E N T _ L I S T _ D I R } / t e s t / t e s t _ c o m m o n . i m p l . h p p "
F I L E S S r c $ { p e r f _ s r c s }
F I L E S I n c l u d e $ { p e r f _ h d r s }
)
ocv_option ( OPENCV_DNN_PERF_CAFFE "Add performance tests of Caffe framework" OFF )
ocv_option ( OPENCV_DNN_PERF_CLCAFFE "Add performance tests of clCaffe framework" OFF )
if ( BUILD_PERF_TESTS )
if ( OPENCV_DNN_PERF_CAFFE
O R $ { t h e _ m o d u l e } _ P E R F _ C A F F E # compatibility for deprecated option
)
find_package ( Caffe QUIET )
if ( Caffe_FOUND )
ocv_target_compile_definitions ( opencv_perf_dnn PRIVATE "HAVE_CAFFE=1" )
ocv_target_link_libraries ( opencv_perf_dnn caffe )
endif ( )
elseif ( OPENCV_DNN_PERF_CLCAFFE
O R $ { t h e _ m o d u l e } _ P E R F _ C A F F E # compatibility for deprecated option
)
find_package ( Caffe QUIET )
if ( Caffe_FOUND )
ocv_target_compile_definitions ( opencv_perf_dnn PRIVATE "HAVE_CLCAFFE=1" )
ocv_target_link_libraries ( opencv_perf_dnn caffe )
endif ( )
endif ( )
endif ( )
if ( DNN_ENABLE_PLUGINS )
ocv_target_compile_definitions ( ${ the_module } PRIVATE ENABLE_PLUGINS )
if ( TARGET opencv_test_dnn )
ocv_target_compile_definitions ( opencv_test_dnn PRIVATE ENABLE_PLUGINS )
endif ( )
if ( OPENCV_DEBUG_POSTFIX )
ocv_append_source_file_compile_definitions ( "${CMAKE_CURRENT_LIST_DIR}/src/backend.cpp" "DEBUG_POSTFIX=${OPENCV_DEBUG_POSTFIX}" )
endif ( )
endif ( )
ocv_option ( OPENCV_TEST_DNN_OPENVINO "Build test with OpenVINO code" ( TARGET ocv.3rdparty.openvino ) )
if ( TARGET ocv.3rdparty.openvino AND OPENCV_TEST_DNN_OPENVINO )
if ( TARGET opencv_test_dnn )
ocv_target_link_libraries ( opencv_test_dnn ocv.3rdparty.openvino )
endif ( )
endif ( )
ocv_option ( OPENCV_TEST_DNN_CANN "Build test with CANN" ( TARGET ocv.3rdparty.cann ) )
if ( TARGET ocv.3rdparty.cann AND OPENCV_TEST_DNN_CANN )
if ( TARGET opencv_test_dnn )
ocv_target_link_libraries ( opencv_test_dnn ocv.3rdparty.cann )
endif ( )
endif ( )
ocv_option ( OPENCV_TEST_DNN_TIMVX "Build test with TIM-VX" ( HAVE_TIMVX ) )
if ( OPENCV_TEST_DNN_TIMVX )
if ( TARGET opencv_test_dnn )
ocv_target_compile_definitions ( opencv_test_dnn PRIVATE "HAVE_TIMVX=1" )
endif ( )
endif ( )
ocv_option ( OPENCV_TEST_DNN_TFLITE "Build test with TFLite" ( OPENCV_DNN_TFLITE ) )
if ( OPENCV_TEST_DNN_TFLITE )
if ( TARGET opencv_test_dnn )
ocv_target_compile_definitions ( opencv_test_dnn PRIVATE "OPENCV_TEST_DNN_TFLITE=1" )
endif ( )
if ( TARGET opencv_perf_dnn )
ocv_target_compile_definitions ( opencv_perf_dnn PRIVATE "OPENCV_TEST_DNN_TFLITE=1" )
endif ( )
endif ( )
