Pinned Tweet
Great news from #CVPR2024 🎉🎉🎉
Happy to share that our paper ElasticDiffusion: Training-free Arbitrary Size Image Generation was accepted @CVPR.
Big thanks to my collaborators @bluevincent and Guha Balakrishnan.
Checkout more details from here: elasticdiffusion.github.io/
MOAYED HAJi ALi retweeted
Happy to announce AlignVLM📏: a novel approach to bridging vision and language latent spaces for multimodal understanding in VLMs! 🌍📄🖼️
🔗 Read the paper: arxiv.org/abs/2502.01341
🧵👇 Thread
MOAYED HAJi ALi retweeted
Introducing
⚗️ Video Alchemist
Our new video model supporting
👪 Multi-subject open-set personalization
🏞️ Foreground & background personalization
🚀 Without the need of inference-time tuning
snap-research.github.io/open…
[Results]
1. Sora girl rides a dinosaur on a savanna
🧵👇
MOAYED HAJi ALi retweeted
Diffusion models are very strong and robust feature extractors, but recent works were only using them for recognition tasks. In our recent work (led by @MoayedHaji), we harness them for video2audio generation: they by far outperform conventional video feature extractors for audio/video temporal alignment and allow to achieve SotA results in sound quality as well: snap-research.github.io/AVLi…
1/6 Introducing AV-Link, an approach to connect video and audio diffusion models in a self-contained framework to enable video-to-audio and audio-to-video generation with superb audio-video synchronization.
Project page: snap-research.github.io/AVLi…
MOAYED HAJi ALi retweeted
Can we use the intermediate representations of pretrained video generation models to generate audio, and vice versa? 🤔🔀
It turns out that this approach can be powerful.
Introducing AV-Link, an approach to connect video and audio diffusion models in a self-contained framework
1/6 Introducing AV-Link, an approach to connect video and audio diffusion models in a self-contained framework to enable video-to-audio and audio-to-video generation with superb audio-video synchronization.
Project page: snap-research.github.io/AVLi…
5/6 For the first time, our approach shows audio-synchronized video generation for “in-the-wild” scenarios, surpassing existing methods in both generation quality and semantic and temporal alignment (snap-research.github.io/AVLi…).
6/6 Check out qualitative examples in the project website (snap-research.github.io/AVLi…). and our preprint (arxiv.org/abs/2412.15191) for more details about the architecture.
MOAYED HAJi ALi retweeted
🧵 1/4 Introducing ColFlor: An Efficient, OCR-Free Vision-Language Document Retrieval Model 🌟
Earlier this year, ColPali transformed document retrieval by removing error-prone OCR pipelines, creating embeddings directly from images. However, its 3 billion parameters make it computationally expensive. That’s where ColFlor comes in! ⚡👇 #AI #NLP
huggingface.co/blog/ahmed-ma…
MOAYED HAJi ALi retweeted
I am excited to share that two of our research works will be presented at ECCV 2024. #ECCV2024 They focus on augmenting language models with fine-grained visual recognition ability.
AutoVER made successful attempts at generative visual recognition. It was accepted to the ECCV 2024 main conference and was invited to the ILR Workshop as an oral presentation. Collaboration w/ @pcascanteb @vislang #Microsoft
Extractive Reranker was accepted to the ILR Workshop as a poster. We explored how the long-context sequence modeling ability of language models can benefit image retrieval, a fundamental computer vision problem.
MOAYED HAJi ALi retweeted
📢Excited to share our recent work ✂️CLIPAway, a plug-and-play solution to simple object removal using latent diffusion models. It works like a charm!
📄Paper: arxiv.org/abs/2406.09368
🔗Project website: yigitekin.github.io/CLIPAway…
🤗@huggingface Spaces Demo: huggingface.co/spaces/yigite…
Introducing CLIPAway🚀
A method for object removal that uses CLIP embeddings to distinguish foreground from background, ensuring accurate inpainting without hallucinations.
Colloboration with @aysegl_dndr @aykuterdemml @erkuterdem @_abyildirim
Check out: yigitekin.github.io/CLIPAway…
MOAYED HAJi ALi retweeted
🎉Exciting News! Introducing SonicDiffusion: A new approach to audio-driven image generation and editing with pretrained diffusion models. Our latest research paper is now available on arXiv!🚀
📄Paper: arxiv.org/abs/2405.00878
🌐Project Page: cyberiada.github.io/SonicDif… [1/8]
MOAYED HAJi ALi retweeted
Hippocrates is finally out! This started as a course project but quickly turned into one of the most satisfying and fun projects we did within our project group. As Emre Can states, we have a lot to explore, while our core team keeps on growing. Stay tuned for updates.
🎉 Exciting news in #HealthcareAI! We're thrilled to introduce Hippocrates, an open-source LLM framework tailored for the medical field! 🚀 #AI #NLProc (1/9)
MOAYED HAJi ALi retweeted
I am reaching out to you to support a crucial cause: helping my family members who are in dire need to find safety and security away from the hardships they're facing in Gaza.
1/5🧵
#Gaza #GazaIsStarving #Palestine #Help #GazaStrip #fund #gofundmedonations #support #diabetes
What is so sensitive about me celebrating my paper?
It severely limited the exposure of my tweet to almost no views
Project webpage: elasticdiffusion.github.io
Arxiv: arxiv.org/abs/2311.18822
Code: github.com/MoayedHajiAli/Ela…
See you in Seattle!