Implement a simple vgg16 network architecture only with forward pass
$ make
g++ -std=c++11 -g -Wall -c -o .out/main.o src/main.cpp
g++ -std=c++11 -g -Wall -c -o .out/network.o src/network.cpp
g++ -std=c++11 -g -Wall -c -o .out/utils.o src/utils.cpp
g++ -std=c++11 -g -Wall -c -o .out/vgg16.o src/vgg16.cpp
g++ -std=c++11 -g -Wall -o vgg16 .out/main.o .out/network.o .out/utils.o .out/vgg16.o
$ make run$ make run
./vgg16
Layer Memory size Param # MAC #
===========================================================================
conv1_1 1*224*224*3 = 150528 1792 86704128
conv1_2 1*224*224*64 = 3211264 36928 1849688064
max_pooling_1 1*224*224*64 = 3211264 0
conv2_1 1*112*112*64 = 802816 73856 924844032
conv2_2 1*112*112*128 = 1605632 147584 1849688064
max_pooling_2 1*112*112*128 = 1605632 0
conv3_1 1*56*56*128 = 401408 295168 924844032
conv3_2 1*56*56*256 = 802816 590080 1849688064
conv3_3 1*56*56*256 = 802816 590080 1849688064
max_pooling_3 1*56*56*256 = 802816 0
conv4_1 1*28*28*256 = 200704 1180160 924844032
conv4_2 1*28*28*512 = 401408 2359808 1849688064
conv4_3 1*28*28*512 = 401408 2359808 1849688064
max_pooling_4 1*28*28*512 = 401408 0
conv5_1 1*14*14*512 = 100352 2359808 462422016
conv5_2 1*14*14*512 = 100352 2359808 462422016
conv5_3 1*14*14*512 = 100352 2359808 462422016
max_pooling_5 1*14*14*512 = 100352 0
fc1_4096 1*1*1*25088 = 25088 102764544 16777216
fc2_4096 1*1*1*4096 = 4096 16781312 16777216
fc3_1000 1*1*1*4096 = 4096 4097000 1000000
===========================================================================