Physical Attribute Prediction

CelebFaces Attributes dataset contains 202,599 face images of the size 178×218 from 10,177 celebrities, each annotated with 40 binary labels indicating facial attributes like hair color, gender and age. Source: Show, Attend and Translate: Unpaired Multi-Domain Image-to-Image Translation with Visual Attention Image Source: http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html

CAR contains visual attributes for objects in the Cityscapes dataset. For each object in an image, we have a list of attributes that depend on the category of the object. For instance, a vehicle category has a visibility attribute while a pedestrian has an activity attribute (walking, standing, etc.). The objective of this dataset is to ease the development of better algorithms for self-driving vehicles as that requires a complete understanding of the entire scene with all of its details including attributes of all objects. We chose Cityscapes as it already contains different types of useful annotations and adding attributes to that will remove a huge burden over developing algorithms with self or semi supervision.

FAD is a dataset that have roughly 200,000 attribute labels for the above traits, for over 10,000 facial images. Source: Predicting Personal Traits from Facial Images using Convolutional Neural Networks Augmented with Facial Landmark Information

We collect a dataset of 805 clean videos that show the action of pouring water in a container. Our dataset spans over 50 unique containers made of 5 different materials, 4 different shapes and with hot and cold water. The dataset can be used for the following tasks: - Physical property understanding (e.g., predict size of container just from the sound of pouring) - Audio generation (given video of pouring, generate a synchronised sound of pouring) - Video generation