With Deep Learning, Computers See Images More Like Humans Do

Scientists asked: Can deep learning distinguish subtle differences between images more like humans do? The answer: Yes. Even self-supervised neural networks work better than prior algorithms designed for this purpose.

AI Can Generate Realistic Sound for Video Clips

Researchers used neural networks to create realistic sound for videos from scratch. The system’s soundtracks fooled humans into thinking they were real.

Researcher Spotlight: Balaji Vasan Srinivasan

Senior research scientist Balaji Vasan Srinivasan was the first external hire at Adobe Research’s Bangalore lab in 2011. He focuses on enabling computers to generate content and create layouts using artificial intelligence. We asked him about his career path, his work training machines to write and design content, and what...

Tap into AI to Improve Video Ads

Can AI help marketers improve video ads, increasing their effectiveness with specific audiences? Adobe Research scientists asked this big question—and have designed a novel technology to answer it.

Adobe Research Papers Featured in Upcoming Conferences

Adobe has achieved a remarkable acceptance rate for research papers at several highly-selective conferences this spring. These conferences serve as key platforms for sharing breakthroughs with the research community.

Sonic Scape: Visualizing Sound for 360 Video

Sonic Scape helps 360 video editors “see” ambisonic sounds in a video using color particles, enabling a unique style of visual editing of the sound’s origin in space and time.

Puppetron: Your Face as a Stylized Work of Art

If you’ve ever wondered what you might look like as a painting by a famous artist, Puppetron has the answer. This experimental technology by Adobe Research and collaborators turns your image into a work of art.

A Papier-Mâché Approach to Learning 3D Surface Generation

Computers have made significant progress in understanding what is depicted in images. How about for 3D? We live in a 3D world. This research project presents a novel approach for generating 3D polygonal meshes.

The Dataset that Made the Cover

Qingnan Zhou, a research engineer at Adobe Research, and a collaborator created a huge poster-style, full-color image showing the objects from their 3D dataset. Now, they are thrilled that their work is being featured on the cover of the journal Computer Graphics Forum.