Training data quality is critical for machine learning model performance. Ideally, quality assurance is an automated process that operates continuously throughout your training data development and improvement processes. Labelbox launched Benchmarks, aka Gold Standard, so that you can do just that!

Benchmarks calculates a numerical value for accuracy by measuring how close a label is to the Benchmark, or 'Ground Truth', label. Benchmarks data is a small subset of the training data labeled by the knowledge expert or data scientist. Benchmarks are automatically distributed to labelers to evaluate their understanding of the labeling task and their accuracy performing it. Benchmarks allows you to track accuracy by overall project, by labelers, and by labels.

Tip: Systemic poor labeler performance is often indicative of poor instructions, while poor performance on certain pieces of data is often indicative of edge cases. Use these values to help you improve your on-boarding and education processes.

Creating training data is often one of the most expensive components of building a machine learning application. Properly monitoring training data quality increases the chance of having a performant model the first time around. And, getting labels right the first time (first pass quality) is far cheaper than the cost of discovering and redoing work to fix the problem. With world class tooling at your fingertips, you can ensure your labeling maintains the level of quality you need to get the modeling results you want.Check out the Benchmarks docs for more information.

Curating and improving your training data involves searching and filtering on what labels exist so that you can find images with specific labeled classes. Filtering labels allows you to streamline actions, such as re-enqueue, delete, and review labels. Click our Label Contains Search support docs for more details.

Screen Shot 2019-01-14 at 9.47.40 AM.png

Asset Info

The asset info feature overlays multi-media metadata so you can get image, text, and video information on a single interface! For instructions on how to add asset info API, click here.


API Key Management

Manage your API keys on Labelbox for ultimate control over systems integration. Labelbox offers multiple API keys per user. Account admins can generate, name, and remove to provide, restrict, and limit access to systems. Typically users integrate with data collection, storage, and management systems. Having unique keys for each system improves security and makes it easier to pin point and resolve issues when they occur.

Visit our API-key support documentation to learn how to set up keys for your project.

Bulk Import

Import data rows via API using JSON/CSV in bulk instead of one-by-one to make imports orders of magnitude faster. Labelbox supports row by row and bulk import. With bulk import you can immediately upload all of your stored data at once. Row by row data enables you to upload data in conjunction with your data collection so collection and annotation can happen in real time.

For specific how-to details visit our data-import support documentation. This is also helpful to read if you want to do row-by-row imports.