Swift for TensorFlow Task 4: Build an Image Classification Model

TensorFlow

This task will help you to become familiar with Swift for TensorFlow so you will then be able to complete more complex tasks. You will learn about using datasets and processing images as input. You will also be able to learn some computer vision principles.

Subtasks

Build and Train a Neural Network that can classify images passed as an image input.
You have to use "Swift for TensorFlow" for building the network.
Use CIFAR-10 dataset for classification.
[OPTIONAL/BONUS] Use ImageNet dataset for classification. Your model should be able to classify at least 10 classes. Searching for a subset of ImageNet can be useful.

Learning Resources

Swift for TensorFlow Models has great documentation and models.
Check out some resources that were built during GSoC '19 (https://github.com/Ayush517/S4TF-Tutorials).

Send us a shared accessible link of your working Colab Project.

Task tags

tensorflow
computer-vision
image-classification
s4tf
swift

Students who completed this task

wallaby123, anigasan, Phoom, Rick Wierenga, Rachin, bkkaggle

Task type

Code
Documentation / Training