Thrilled to share EmerDiff, accepted at #ICLR2024 ! EmerDiff is an unsupervised image segmentor solely built on the semantic knowledge extracted from a pre-trained diffusion model. Joint work w/ Amirmojtaba Sabour Sanja Fidler Seung Wook Kim Webpage: kmcode1.github.io/Projects/EmerD…

Excited to release this work on exploring unsupervised segmentation with pretrained image diffusion models, awesome work led by Koichi Namekata !

I'll be presenting our #ICLR2024 poster "EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion Models" (#83, Wed morning), fine-grained unsupervised image segmentation with DMs. Feel free to drop by if you are interested in exploring the semantic understanding of DMs!

Thrilled to announce that SG-I2V has been accepted at #ICLR2025 ! Huge thanks to the collaborators, reviewers, and ACs. Looking forward to presenting this in Singapore!

I'll be presenting our #ICLR2025 poster "SG-I2V: Self-Guided Trajectory Control in Image-to-Video Generation" (#189, Thu afternoon), trajectory conditioned i2v generation w/o fine-tuning. Feel free to drop by if you are interested in exploring the zero-shot capabilities of VDMs!