Bnotate creates a more natural way to annotate images for training computer vision and artificial intelligence models. The annotation methodology follows “Extreme clicking for efficient object annotation” by Papadopolous et al. (ICCV 2017) where one simply touches the four extreme points of the object, i.e. points belonging to the top, bottom, left-most, and right-most parts of the object. Once complete, the annotations are exported into a convenient CSV format similar to the bounding boxes format for the Open Images Dataset V5.
Key functionality:
– Download image files from cloud data stores to the device for use in a offline mode of operation
– Pinch to zoom (up to 3x) and press and drag to pan for easy image manipulation before locking in an annotation
– Export format is similar to expectations for many pre-trained neural networks and cloud computer vision platforms