# Workflow We propose the following workflow to create Deep Learning inspection tools. **Deep Learning Studio** implements this workflow through the creation of a project and its tabs. The API is also designed to support this workflow. Which tool should I use for my application? Manage the dataset and annotate your images. Split your dataset in 3 parts for robust evaluation of performances. Select and configure data augmentation settings for your data. Training a Deep Learning tool. Using a Deep Learning tool and making prediction on new images. Benchmark the speed of your trained tools and optimize the execution settings to maximize inference speed.