Artificial Intelligence && Deep Learning
57.7K subscribers
166 photos
21 videos
57 files
741 links
Channel for who have a passion for -
* Artificial Intelligence
* Machine Learning
* Deep Learning
* Data Science
* Computer vision
* Image Processing
* Research Papers

With advertising offers contact: @ai_adminn
Download Telegram
Want to jump ahead in artificial intelligence and/or digital pathology? Excited to share that after 2+ years of development PathML 2.0 is out! An open source #computational #pathology software library created by Dana-Farber Cancer Institute/Harvard Medical School and Weill Cornell Medicine led by Massimo Loda to lower the barrier to entry to #digitalpathology and #artificialintelligence , and streamline all #imageanalysis or #deeplearning workflows.

Code: https://github.com/Dana-Farber-AIOS/pathml
This media is not supported in your browser
VIEW IN TELEGRAM
🆔🆔 Magic-Me: Identity-Specific Video 🆔🆔

👉hashtag#ByteDance (+UC Berkeley) unveils VCD for video-gen: with just a few images of a specific identity it can generate temporal consistent videos aligned with the given prompt. Impressive results, source code under Apache 2.0 💙

𝐇𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬:
Novel Video Custom Diffusion (VCD) framework
High-Quality ID-specific videos generation
Improvement in aligning IDs-images and text
Robust 3D Gaussian Noise Prior for denoising
Better Inter-frame correlation / video consistency
New modules F-VCD/T-VCD for videos upscale
New train with masked loss by prompt-to-segmentation

hashtag#artificialintelligence hashtag#machinelearning hashtag#ml hashtag#AI hashtag#deeplearning hashtag#computervision hashtag#AIwithPapers hashtag#metaverse

👉Channel: @deeplearning_ai
👉Paper https://arxiv.org/pdf/2402.09368.pdf
👉Project https://magic-me-webpage.github.io/
👉Code https://github.com/Zhen-Dong/Magic-Me
This media is not supported in your browser
VIEW IN TELEGRAM
Introducing ECoDepth: The New Benchmark in Diffusive Mono-Depth

From the labs of IITD, we unveil ECoDepth - our groundbreaking SIDE model powered by a diffusion backbone and enriched with ViT embeddings. This innovation sets a new standard in single image depth estimation (SIDE), offering unprecedented accuracy and semantic understanding.

Key Features:

Revolutionary MDE approach tailored for SIDE tasks
Enhanced semantic context via ViT embeddings
Superior performance in zero-shot transfer tasks
Surpasses previous SOTA models by up to 14%

Dive into the future of depth estimation with ECoDepth. Access our source code and explore the full potential of our model.

📖 Read the Paper
💻 Get the Code

#ArtificialIntelligence #MachineLearning #DeepLearning #ComputerVision #AIwithPapers #Metaverse

join our community:
👉 @deeplearning_ai
Media is too big
VIEW IN TELEGRAM
Neural Bodies with Clothes: Overview

Introduction:
Neural-ABC, a cutting-edge parametric model developed by the University of Science & Technology of China, innovatively represents clothed human bodies.

Key Features:

Novel approach for modeling clothed human figures.
Unified framework accommodating various clothing types.
Consistent representation of both body and clothing.
Enables seamless modification of identity, shape, clothing, and pose.
Extensive dataset with detailed clothing information.

Explore More:
💻Project Details: Discover More
📖Read the Paper: Access Here
💻Source Code: Explore on GitHub

Relevance: #artificialintelligence #machinelearning #AI #deeplearning #computervision

join our community:
👉 @deeplearning_ai