Google's Cloud AutoML Vision is a one-click training solution for object detection models. It uses its own CSV format that references image files in Google Cloud Storage buckets. To learn more, check out our full benchmark of Google Cloud AutoML Vision.
With Roboflow Pro, you can export annotations from any format into the proper format to get up and running with Cloud AutoML in seconds.
Below, learn the structure of Google Cloud AutoML Vision CSV.
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/18x5vcUUHj2YgtdY6wkH/transformed.jpg,helmet,0.256,0.224,0.466,0.224,0.466,0.36666666666666664,0.256,0.36666666666666664
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/18x5vcUUHj2YgtdY6wkH/transformed.jpg,helmet,0.316,0.6333333333333333,0.58,0.6333333333333333,0.58,0.8013333333333333,0.316,0.8013333333333333
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/3xEsBTofg5wE5QlciVbL/transformed.jpg,helmet,0.09183673469387756,0.007272727272727273,0.2653061224489796,0.007272727272727273,0.2653061224489796,0.31636363636363635,0.09183673469387756,0.31636363636363635
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/3xEsBTofg5wE5QlciVbL/transformed.jpg,helmet,0.6612244897959184,0.10545454545454545,0.8081632653061225,0.10545454545454545,0.8081632653061225,0.4,0.6612244897959184,0.4
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/3xEsBTofg5wE5QlciVbL/transformed.jpg,helmet,0.5489795918367347,0.20363636363636364,0.6632653061224489,0.20363636363636364,0.6632653061224489,0.43636363636363634,0.5489795918367347,0.43636363636363634
TRAIN,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/3xEsBTofg5wE5QlciVbL/transformed.jpg,helmet,0.24489795918367346,0.04363636363636364,0.363265306122449,0.04363636363636364,0.363265306122449,0.29818181818181816,0.24489795918367346,0.29818181818181816
VALIDATE,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/1HwmHNcOur7rLexvxKAE/transformed.jpg,helmet,0.868,0.30513595166163143,0.94,0.30513595166163143,0.94,0.4350453172205438,0.868,0.4350453172205438
VALIDATE,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/1HwmHNcOur7rLexvxKAE/transformed.jpg,helmet,0.564,0.3323262839879154,0.636,0.3323262839879154,0.636,0.4229607250755287,0.564,0.4229607250755287
TEST,gs://roboflow-platform-transforms/JW38IllTMkcS3HGC9IdKSwGtc5g1/DpXAgMz4QM6oXS6NhzWm/transformed.jpg,helmet,0.6,0.40168539325842695,0.674,0.40168539325842695,0.674,0.547752808988764,0.6,0.547752808988764With Roboflow supervision, an open source Python package with utilities for completing computer vision tasks, you can merge and split detections in Google Cloud AutoML Vision CSV. Read our dedicated guides to learn how to merge and split Google Cloud AutoML Vision CSV detections.
Below, see model architectures that require data in the Google Cloud AutoML Vision CSV format when training a new model.
On each page below, you can find links to our guides that show how to plot predictions from the model, and complete other common tasks like detecting small objects with the model.