This announcement by Facebook could be a major breakthrough in artificial intelligence (AI). FB AI has brought a self-supervised learning paradigm shift to computer vision. It announced on its blog that it had developed SEER (SElf-supERvised), a new billion-parameter self-supervised computer vision model that can learn from any random group of images on the internet — without the need for careful curation & labeling that goes into most computer vision training today. Which means building a highly accurate unsupervised model at image recognition.
After pretraining on a billion random, unlabeled & “uncurated” public Instagram images, SEER outperformed the most advanced, state-of-the-art self-supervised systems, reaching 84.2% top-1 accuracy on ImageNet. SEER also outperformed state-of-the-art supervised models on downstream tasks, including low-shot, object detection, segmentation, and image classification. When trained with just 10% of the examples in the ImageNet data set, SEER still achieved 77.9% top-1 accuracy on the full data set. When trained with just 1% of the annotated ImageNet examples, SEER achieved 60.5% top-1 accuracy.
Facebook said SEER’s performance demonstrated that self-supervised learning could excel at computer vision tasks in real-world settings.
Facebook’s SEER AI Has Trained Itself to Recognize Objects – petapixel.com
Facebook today announced a major breakthrough: Facebook’s AI, named SEER (SElf-supERvised AI), was able to correctly identify and categorize objects in photos without the aid of humans with a high degree of accuracy.
Yan Lecun, Facebook’s chief artificial intelligence scientist, wants the…
Link: Facebook’s SEER AI Has Trained Itself to Recognize Objects
via petapixel.com
