Meta has created a dataset of first-person point of view videos taken in diverse environments. The dataset contains 12,000 videos featuring 200 object categories. Each video was filmed from a first-person point of view with varying camera angles, lighting conditions, background clutter, and object distances. Researchers can use the dataset to train object classification models.
Image credit: Flickr user jose.designs