This 2-hour-long project-based course will teach you how to build and train a convolutional network (CNN), in Keras. It will be used to recognize facial expressions. The data is 48x48 pixels grayscale images of faces. The objective of the data is to classify each facial expression based on its emotion into one of seven categories (0=Angry or 1=Disgust, 2=Fear and 3=Happy), 4=Sad (Surprise), 5=Sad, 6=Neutral). OpenCV will automatically detect faces in images, and you will draw bounding boxes around them. Once you have saved, trained, and exported CNN, you can directly serve the trained model to the web interface and perform real time facial expression recognition using video and image data. Coursera's hands on project platform Rhyme is where this course is run. Rhyme allows you to work in a browser-based manner on projects. Instant access to pre-configured cloud desktops that contain all the software and data needed for your project will be available. Everything is already configured in your internet browser, so you can focus on learning. You'll have instant access to the cloud desktop with Python, Jupyter and Keras installed for this project. Notes: You can access the cloud desktop five times. However, you can access the instructions videos as many times you wish. This course is best for learners who live in the North America region. We are currently working to offer the same experience in other areas.
