ML Trainer: Make Training Data
DATA COLLECTION MADE EASY!
An easy to use, open source tool designed to help developers record training data for machine learning models!
Creating original training data for Image Classification machine learning models just got a little easier! ML Trainer allows developers to quickly capture and export thousands of images to the Photos app, allowing every image to be imported with iCloud or the built in Image Capture app on Mac.
Press the Scan button to capture a preset amount of images as you move closer to or pivot around your subject, or Tap the Camera button to capture a single picture. Toggle the Flashlight to improve results in low light conditions, and tap the Save button to export any captured images to the Photos app.
While each image is always captured at a speed of 3 Frames Per Second, you can adjust the Frame Count of each Scan in the Settings Menu. A larger Frame Count will save you time, while a lower Frame Count will help improve accuracy across different angles. Enabling Crosshairs and Guides in the Settings Menu can also help improve accuracy.
This app was specifically designed to speed up the process of importing data into the Xcode Developer Tool named CreateML. Exported images should also be compatible with other platforms like TensorFlow and Azure Machine Learning. A wired connection to a macOS device with the Image Capture app open will always be the fastest way to import your data to desktop.
Privacy Policy
ML Trainer presents a unique user experience allowing the camera’s to interact with the world around you to position the game world. Camera access can be toggled on or off at any time in your device's settings. In order for augmented reality or virtual reality to be possible, this app requires access to your device's camera.
This app does not store user input or collect any personal identifiable information unless you make an In App Purchase. This app does not use any Third Party API’s, and does not share any information with any third party advertiser or entity. The camera images and resulting depth data are only used for gameplay purposes. Camera access is required to collect contextualized depth data from the API's provided by Apple. Any collected data is stored on the device only for the duration of the current game session. When you take a photo or video using the 3D Camera, the resulting exported file does not include depth data and only portrays the results.
User data is never transmitted, tracked, targeted, or stored when advertisements are displayed. All advertisement URL’s are stored locally in the applications source code, and randomly distributed at a set frequency. Advertised content will only contain family appropriate applications developed by Near Future Marketing Inc.
Every time the application is closed, all tracking data is deleted. World data saved by the user is always stored locally, and can only be shared when a multiplayer session is initiated by the user. Face data is never stored remotely, given to third parties, or used for any non-gameplay/entertainment purposes at any time.
If you have any additional questions, please reach out to us via email: privacy@nearfuture.marketing
Thank you for using ML Trainer, enjoy!