Roboflow Changelog: December 2020
Blog post from Roboflow
Roboflow's December 2020 changelog highlights various updates and improvements across its product offerings, including Roboflow Organize, Train, Infer, and Annotate. Key enhancements include the ability to rebalance train/test splits, faster thumbnail generation, and new annotation formats like SuperAnnotate JSON, Kaggle NFL CSV, and PKLot XML. The Roboflow Train feature now provides class-specific test set metrics and supports upcoming on-device inference, while Roboflow Infer has improved speed, reliability, and introduced a video inference script. The company also announced the private beta testing of an on-device offering and imminent wide release of a self-serve labeling tool. Additionally, new industry pages for Agriculture and Safety & Security were launched, along with updates to the enterprise and pricing pages. Roboflow's advancements and contributions were recognized by mentions in Augmented Startups, Machine Learning Monthly, and a citation in Nature.