我正在使用google cloud run来部署我的应用程序。每次我部署应用程序时，构建大约需要22分钟，因为dlib库（必需库之一）非常大。我最终得到了如下的结果：

Building wheels for collected packages: dlib, face-recognition-models
  Building wheel for dlib (setup.py): started
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...
  Building wheel for dlib (setup.py): still running...

有没有办法缓存docker图像，这样我就可以加快构建速度？无论何时我在本地运行pip install -r requirements.txt，它都非常快，因为pip在它的缓存中有库。我在Artifact Registry中保存了以前构建的docker图像。