Cyberiad Research Group
Teaching machines to see, imagine, and reason about the visual world.
We develop generative and multimodal AI systems that understand, create, and edit images and video—bringing together computer vision, machine learning, graphics, language, and signal processing.
Based at Koç University and Hacettepe University · Affiliated with KUIS AI
What we study
Research themes
01
Generative visual intelligence
Controllable image and video creation, editing, motion planning, and neural scene representations.
02Multimodal reasoning
Vision-language models, compositional generalization, temporal grounding, and behavioral evaluation.
03Structured video
Continuous, compact representations for coherent motion, reconstruction, resampling, and editing.
04Spherical visual intelligence
Geometry-aware 360° perception, visual attention, eye tracking, and spatial audio.
Latest work
Featured research
IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)
Spherical Vision Transformers for Audio-Visual Saliency Prediction in 360-Degree Videos
We introduce the YT360-EyeTracking dataset and spherical geometry-aware vision transformers that combine visual and spatial-audio cues for saliency prediction in 360-degree video.
Our group
Research across institutions and disciplines
Cyberiad brings together researchers in computer vision, machine learning, graphics, language, and signal processing at Koç University and Hacettepe University.
Meet the team →
Work with us
Help shape the next generation of AI.
We welcome ambitious MSc and PhD students, postdoctoral researchers, research assistants, and visiting students.
