Subject Lifting is now available through Vision and VisionKit in iOS17!! Can't wait to see how these APIs will be leveraged to build delightful experiences. If this is of interest, do check out this presentation at WWDC. https://lnkd.in/e6zegC3w #segmentation #ios17 #wwdc23 #ipados17 #macOSSonoma
Vignesh Jagadeesh’s Post
More Relevant Posts
-
It was exciting to see Apple Intelligence (https://lnkd.in/gTnsjCEa) announced last week! Thankful to be working alongside fantastic teams at Apple on Search, Genmoji, Image Wand, Image Playground, Memory Creation, and Writing Tools. Multimodal Vision-Language models viewed as a constituent of Apple Intelligence provide for some distinct and unique applied research opportunities … across pre-training, safety, alignment, adaptation, performance, privacy, PCC and on-device deployment. If these sound like areas you are passionate about, please reach out. We are hiring! (https://lnkd.in/gDUAWGPM) #ios18 #appleintelligence #genmoji #macOSSequoia #AppleNeuralEngine #generativeAI
-
Vignesh Jagadeesh reposted this
We share how Apple applies differential privacy and secure aggregation to learn about the kinds of photos people take at frequently visited locations (iconic scenes) without personally identifiable data leaving their device. This approach is used in several features within the Photos App, including choosing key photos for Memories, and selecting key photos for locations in the Places Map. Check out our research article on our ML Research site, "Learning Iconic Scenes with Differential Privacy" here: https://lnkd.in/gqu2nm4a
Learning Iconic Scenes with Differential Privacy
machinelearning.apple.com
-
A recent blogpost discussing image segmentation technology backing subject lifting and stickers in iOS17. https://lnkd.in/gQCHhFjW #ios17 #ipados17 #segmentation #inference #stickers #macossonoma
Fast Class-Agnostic Salient Object Segmentation
machinelearning.apple.com
-
Some really nice work and updates on running Stable Diffusion in Apple Silicon. Congratulations to all the contributors 👏👏 #stablediffusion #applesilicon #ios17 #macossonoma
Exciting updates to Stable Diffusion with Core ML on Apple Silicon! - 6-bit weight compression that yields just under 1 GB - Up to 30% improved Neural Engine performance - New benchmarks on iPhone, iPad and Macs - Multilingual system text encoder support - ControlNet Credits go to our open-source contributors as well as the hard-working teams at Apple that delivered great core technologies and developer tools at WWDC23 last week! Resources: - Core ML Stable Diffusion docs: https://lnkd.in/dJTXcwBG - Use Core ML Tools for machine learning model compression: https://lnkd.in/gmnhmJ6p - Explore Natural Language multilingual models: https://lnkd.in/g2nK5t8Q #stablediffusion #applesilicon #deeplearning #machinelearning #wwdc2023 #wwdc23
GitHub - apple/ml-stable-diffusion: Stable Diffusion with Core ML on Apple Silicon
github.com
-
Delighted to see the response live stickers in iOS17 has been receiving! Backing these features are a variety of cutting edge computational blocks that run on device. A few examples include a class agnostic static image segmenter, an efficient and high quality tracker, multimodal ranking / suggestions, image stylization and efficient temporal compression. (https://lnkd.in/gkDGhHkQ) I learnt a lot working with a fantastic cross functional team through the project. As a user, I especially enjoy the systemwide integration of stickers. Thankful as always to be part of a great team at Apple. We are actively hiring for Multimodal ML R&D in Zurich and Cupertino. Please check out the roles, and reach out / apply if you are interested. https://lnkd.in/gW8Nn5Np https://lnkd.in/gtKxjQ8D #ios17 #ipadOS17 #macOSSonoma #segmentation #tracking #ranking #stickers #messages #emoji
iOS 17 Preview
apple.com
-
CreateML now provides access to updated higher quality visual representations for training downstream workflows. These representations are trained with a multimodal objective and can be efficiently computed on device. https://lnkd.in/e6af4tNZ Our team is actively hiring! If you are interested in Multimodal ML workflows, please reach out or consider applying to https://lnkd.in/eQuiH94J #ios17 #macOS #wwdc23 #embeddings #features #training
Discover machine learning enhancements in Create ML - WWDC23 - Videos - Apple Developer
developer.apple.com
-
Glad that our team contributed towards advancing User Safety in iOS17 through Communication Safety & Sensitive Content Warning. #wwdc23 #ios17 #macOS #safety
Apple announces powerful new privacy and security features
apple.com
-
Vignesh Jagadeesh reposted this
Are you excited about #generativeai? We are looking for experts in this space to join our applied ML R&D team at Apple! You will be inventing and shipping the next generation of these core technologies with a focused team. Here is the link to apply: https://lnkd.in/gSfYSrP5 Here is a piece of our early work in this space: https://lnkd.in/dgSDchid #hiring #apple #machinelearning #deeplearning #stablediffusion
-
Vignesh Jagadeesh reposted this
Delighted to share #stablediffusion with Core ML on Apple Silicon built on top of Hugging Face diffusers! Today's release of macOS Ventura 13.1 Beta 4 and iOS & iPadOS 16.2 Beta 4 include optimizations that let Stable Diffusion run with improved efficiency on the Apple Neural Engine as well as on Apple Silicon GPU. We share sample code for model conversion from PyTorch to Core ML and have example Python pipelines for text-to-image using Core ML models run with coremltools and diffusers. We also released a Swift package in order to make it easier for developers deploy this technology in their apps. As a highlight, the baseline configuration of M2 MacBook Air with 8GB RAM runs https://lnkd.in/dtaf4gkk for 50 iterations in 18 seconds. Distilled Stable Diffusion 2, which was announced recently (https://lnkd.in/dTYf59q2), requires 1 to 4 iterations instead of 50. The same M2 device should generate an image in <<1 second. Blog Post: https://lnkd.in/dgSDchid Code: https://lnkd.in/dJTXcwBG #stablediffusion #generativeai #apple #deeplearning
-