Image Recognition Mini-Project Walkthrough - [2020-09-01]

For office hours, we do a walkthrough of an image recognition mini-project. Using Google Colab, we go over a typical process of getting data onto the notebook, creating a baseline, training a CNN/deep learning model, and evaluating our models.

0:00 Alright & Welcome
0:32 Review of Today
2:40 CNN Lesson Example
4:55 Google Colab Intro
7:30 Using Colab: Connecting GPU Runtime
10:00 Using Colab: Terminal Commands (pip install)
11:09 Using Colab: Loading Data
18:30 Running this project
19:42 Initial Exploration
27:10 Building a Baseline Model with Pixel Brightness
31:54 Building a Deep Learning Model
32:12 Why We One-Hot Encode Labels
35:11 Explaining Our Model Architectures
35:45 Using TensorFlow Backend
47:00 Evaluating Model History
52:15 Keras Model Checkpoints (Early Stopping)
1:02:29 Training Multiple Models
1:05:09 Questions?