VisionTransformer or CNN what should I choose?

I'm a final year undergraduate student, currently working on my final year project. My project is multiclass image classification. I want to make a full working web application for final presentation. Should I use vision transformer or CNN? Any suggestions?

Also I've used TensorFlow before but now I wamt to switch to PyTorch. Help me out if you have any resources for begginer to deployment level PyTorch code 🙏

application classification cnn deeplearning image presentation project pytorch tensorflow transformer undergraduate vision web