A CNN-based face recognition project targeted at the ORL dataset
Installing dependency libraries: pip install -r requirements.txt
Unzipping the dataset.
Starting the model training program: python face_recognition.py
During training, sometimes the random division of the dataset may lead to poor convergence of the loss value.This may be due to the small amount of data.
The accuracy rate is between 93% and 100% on the test set.
这是一个基于卷积神经网络针对于ORL数据集的人脸识别项目
安装依赖模块: pip install -r requirements.txt
解压数据集。
启动训练模型程序: python face_recognition.py
可能是数据量较小的原因,在训练时,由于数据集的随机划分,有时会导致损失值的不收敛。
在测试集上准确率在93%~100%。