激光点云感知 voxnet本质

voxnet 的本质就是三维卷积加 softmax 多分类而已

model = Sequential()
model.add(Reshape((30, 30, 30, 1), input_shape=(30, 30, 30)))  # 1 in-channel
model.add(Conv3D(16, 6, strides=2, activation='relu'))
model.add(Conv3D(64, 5, strides=2, activation='relu'))
model.add(Conv3D(64, 5, strides=2, activation='relu'))
model.add(Flatten())
model.add(Dense(10, activation='softmax'))

​​keras 版本​​​​keras​​

​​古老的tensorflow​​​​论文​​