Convolutional neural nework (CNN) is a type of deep neural network that fully extends what you learned about deep neural network that allows us to identify multiple objects within an image. We have learned with MNIST that deep learning can be used to identify handwriting, but that’s only on the basis of a singular entity.
In the world of applied science, we can simply leverage existing cloud computing services such as AWS Rekognition to identify objects such as people and things. There isn’t a need to train a model from scratch as it is less expensive and time efficient when cloud providers have such a service for us to leverage.
Nevertheless, you’ll still need to model the data and output based on your use cases.
Exploring CNN
Let’s open up the file(s) in the 01-Par_Explore_CNNs folder to get started.
The school has provided an article on a high-level explanation of a CNN: https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/
Pretrained Models
Let’s open up the file(s) in the 02-Ins_PreTrained_Models folder to get started.
Very similar to using pre-built models such as chatbots and generative AI (Llama3 and ChatGPT), we can use pre-built models to lower the barrier of entry to identify objects within an image: https://keras.io/api/applications/vgg/#vgg19-function
Students Do: Xception
Let’s open up the file(s) in the 03-Stu_Xception folder to get started.
