Hope you've been enjoying summer!
Announcing the Masterful CLI Trainer
Low-code computer vision development: The recently released Masterful CLI Trainer takes care of all the work to get from labeled images to a trained model. No coding required.
Pass in a short YAML config file, labels as a csv, and images in folders.
The CLI returns a trained model on disk after training and printing out some evaluation metrics.
Don’t write boilerplate code. The Masterful CLI Trainer handles all the work of data loading, evaluation, console output, and building an inference model.
Powerful algorithms: The CLI Trainer is built on the Masterful Python API, so you get all the power of the platform to create production-grade models:
- dynamic regularization to squeeze all the performance out of your labeled data
- semi-supervised learning to improve your models with unlabeled data
- meta-learning to virtually eliminate hyperparameter tuning
- GPU optimization to minimize your cloud compute costs
But... if you love debugging tedious tensor shape mismatches, CPU bottlenecks, image range and type issues, and RGB/BGR channel ordering... then maybe the CLI Trainer isn't for you. 😉
Get started in 10 minutes with the Masterful CLI Trainer docs and blog post.
If you are putting models into production in the cloud, at the edge on Nvidia Jetson devices, with hosted environments like Sagemaker Endpoints, or even setting up your own Kubernetes cluster, reach out if you could use some help or just to trade notes.
We've had recent conversations with Nvidia, Microsoft Azure, AWS, and Google Cloud Compute to understand their latest deployment options.
Supported Model Formats: Speaking of which, the Masterful CLI automatically saves a trained model to disk in both Tensorflow Saved Model and ONNX Formats.
Computer Vision is Everywhere
CIO magazine published an article giving examples of how CV is revolutionizing existing industrial processes and creating entirely new capabilities. CV is on the radar of IT departments , not just engineering teams.
Delivery: Walmart is investing in drones and autonomous vehicles for delivery. Both are built around CV.
Warehouses: Fabric builds fulfillment centers for retailers and uses detection and segmentation models to enable robots to pick and place items in warehouses.
Manufacturing: Insight helped a manufacturer built a custom application to count items in a bin - including items hidden beneath other items.
Other applications: Microsoft and PWC are working to help doctors diagnose a type of lung disease, idiopathic pulmonary fibrosis. The article also mentions hospitality, cybersecurity, agriculture, and autonomous vehicle applications.
Ian Goodfellow officially announced that he’s joining DeepMind after leaving Apple.
Join the Community
We'd love to hear what you are working on! Join our community slack.