Make high dynamic range image from low dynamic range images with fewer artifacts

Train thermal image tracker on plentiful RGB images for better performance

Robust fast tracking using multi-level feature transformer

Improve 3D depth estimates by using relationships between objects

Making blueprints from images of carpentered items

Remove image noise by adjust for that image a neural network trained on other data

Unsupervised segmentation of image into parts

Convert face images into speech waveforms

Aligning 3D points when there are large rotations and translations

Improve resolution of image when noise unknown by training with artificial data

Find moment in video matching a description

Face.evoLVe: Face Recognition Library (align, detect, augment, ResNet, DenseNet, MobileNet, etc.)

YOLOX is best YOLO object detector and has simpler design

Cut out soft foreground in natural image with deep learning

Correcting over-exposed image by combining many images using texture

Get observer movement with less image data more accurately by reinforcement learning

Facial feature for image retrieval and for copying to another image

Tracking flexible surfaces in monocular video

Comparing 7 super-resolution methods for video surveillance

Find defects in part images when only train with defect-free images