Answering spatial questions about scene using image and 3D with ScanQA

Sharper depth from single image by using scene structure and details with CADepth-Net

Accurate facial computer vision with Microsoft system trained only on artificial images

Untrained object detection using over 100 times less data by flexible captioning with OTTER

Survey of methods for detecting new objects with only a few examples

Segment objects in images with arbitrary text queries using OpenSeg

Reidentify people in new scenes better by accounting for camera styles with CA-UReID

Improved deblurring of images with guided image generation in W-DIP

Segment object in an image according to a text description

Improved object segmentation in video by using object descriptors instead of pixel matching

Answer questions about scene using image and 3D information

Get 3D shape, pose, and relative depth of people from a single image despite occlusion

Improved super-resolution for display screens by using transformer designed for screen content

Better enhancement of dim images with edge-awareness using CSDNet

Robot grips new objects in new poses from 10 examples using neural descriptor fields

Recognize 3D objects when only trained on 2D image and text pairs with PointCLIP

Track people in images better by building 3D model from image and predicting appearance

Calibrate cameras from video with sub-pixel error using self-supervision

Restore faces blurred by air turbulence with prior knowledge from GAN network

Detect new objects better by teaching classifier not to ignore unlabeled objects