🦙 Llama-2: the Open-Source "ChatGPT" 🦙

👉GenAI, #Meta unveils Llama-2: a collection of LLMs ranging in scale 7-70B params. Challenging with #chatgpt, but open.

😎Review https://t.ly/bLJgP
😎Paper https://t.ly/AOXru
😎Project https://ai.meta.com/llama

🐘 Controllable Synthetic Data (extending Image-Net) 🐘

👉#META's PUG, a new generation of interactive environments for representation learning. Extending Image-Net!

😎Review https://t.ly/nCYs0
😎Paper arxiv.org/pdf/2308.03977.pdf
😎Project pug.metademolab.com
😎Code github.com/facebookresearch/PUG

⛺FACET: Fairness in Computer Vision⛺

👉#META AI opens a large, publicly available dataset for classification, detection & segmentation. Potential performance disparities & challenges across sensitive demographic attributes

😎Review https://t.ly/mKn-t
😎Paper arxiv.org/pdf/2309.00035.pdf
😎Dataset https://facet.iss.onetademolab.com/

🔥🔥 #META's DINOv2 is now commercial! 🔥🔥

👉Universal features for image classification, instance retrieval, video understanding, depth & semantic segmentation. Now suitable for commercial.

😎Review https://t.ly/LNrGy
😎Paper arxiv.org/pdf/2304.07193.pdf
😎Code github.com/facebookresearch/dinov2
😎Demo dinov2.metademolab.com/

✌️ Relighted 3D Hands 🤞

👉#META unveils Re:InterHand: a large dataset of relighted 3D interacting hands

😎Review https://t.ly/I1dQk
😎Paper arxiv.org/pdf/2310.17768.pdf
😎Project mks0601.github.io/ReInterHand
😎Data github.com/mks0601/ReInterHand

🐓 Emu: image edit / video gen. 🐓

👉#Meta the new SOTA in text-to-video generation and instruction-based image editing

👉 Review https://t.ly/PMTBc
👉 Paper (images): https://lnkd.in/eVadH-QS
👉 Project https://lnkd.in/eG8eWUJY
👉 Paper (video): https://lnkd.in/eVadH-QS
👉 Project https://lnkd.in/eu6Zu6gp

🪖RT Humanoid from Head-Mounted Sensors🪖

👉#META (+CMU) announced SimXR, a method for controlling a simulated avatar from info obtained from AR/VR headsets

👉Review https://t.ly/Si2Mp
👉Paper arxiv.org/pdf/2403.06862.pdf
👉Project www.zhengyiluo.com/SimXR/

🧤HOT3D Hand/Object Tracking🧤

👉#Meta opens a novel egocentric dataset for 3D hand & object tracking. A new benchmark for vision-based understanding of 3D hand-object interactions. Dataset available 💙

👉Review https://t.ly/cD76F
👉Paper https://lnkd.in/e6_7UNny
👉Data https://lnkd.in/e6P-sQFK

🔥🔥 SAM v2 is out! 🔥🔥

👉#Meta announced SAM 2, the novel unified model for real-time promptable segmentation in images and videos. 6x faster, it's the new SOTA by a large margin. Source Code, Dataset, Models & Demo released under permissive licenses💙

👉Review https://t.ly/oovJZ
👉Paper https://t.ly/sCxMY
👉Demo https://sam2.metademolab.com
👉Project ai.meta.com/blog/segment-anything-2/
👉Models github.com/facebookresearch/segment-anything-2

🐏 EFM3D: 3D Ego-Foundation 🐏

👉#META presents EFM3D, the first benchmark for 3D object detection and surface regression on HQ annotated egocentric data of Project Aria. Datasets & Code released💙

👉Review https://t.ly/cDJv6
👉Paper arxiv.org/pdf/2406.10224
👉Project www.projectaria.com/datasets/aeo/
👉Repo github.com/facebookresearch/efm3d

🔥 CoTracker3 by #META is out! 🔥

👉#Meta (+VGG Oxford) unveils CoTracker3, a new tracker that outperforms the previous SoTA by a large margin using only the 0.1% of the training data 🤯🤯🤯

👉Review https://t.ly/TcRIv
👉Paper arxiv.org/pdf/2410.11831
👉Project cotracker3.github.io/
👉Code github.com/facebookresearch/co-tracker

☀️ Universal Relightable Avatars ☀️

👉#Meta unveils URAvatar, photorealistic & relightable avatars from phone scan with unknown illumination. Stunning results!

👉Review https://t.ly/U-ESX
👉Paper arxiv.org/pdf/2410.24223
👉Project junxuan-li.github.io/urgca-website

❤️‍🔥 Uncommon object in #3D ❤️‍🔥

👉#META releases uCO3D, a new object-centric dataset for 3D AI. The largest publicly-available collection of HD videos of objects with 3D annotations that ensures full-360◦ coverage. Code & data under CCA 4.0💙

👉Review https://t.ly/Z_tvA
👉Paper https://arxiv.org/pdf/2501.07574
👉Project https://uco3d.github.io/
👉Repo github.com/facebookresearch/uco3d

☀️ Relightable Full-Body Avatars ☀️

👉#Meta unveils the first approach ever to jointly model the relightable appearance of the body, face, and hands of drivable avatars.

👉Review https://t.ly/kx9gf
👉Paper arxiv.org/pdf/2501.14726
👉Project neuralbodies.github.io/RFGCA

🔥 VideoJAM: #META's Video-Model (SOTA) 🔥

👉#META's VideoJAM: the new SOTA (by large margin) in motion coherence for video generation, much better than SORA! A strong motion prior into any video-gen model. Impressive results, no code announced🥲

👉Review https://shorturl.at/id7Bt
👉Paper https://arxiv.org/pdf/2502.02492
👉Project https://hila-chefer.github.io/videojam-paper.github.io/

🤖 META Human-Robot 🤖

👉#META PARTNR: novel benchmark for Planning And Reasoning Tasks in humaN-Robot collaboration. The largest benchmark of its kind: 100,000+ natural language tasks, spanning 60 houses and 5,819 unique objects. Code & Data (🤗) under MIT💙

👉Review https://t.ly/zcN0K
👉Paper arxiv.org/pdf/2411.00081
👉Repo github.com/facebookresearch/partnr-planner
🤗Data huggingface.co/datasets/ai-habitat/partnr_episodes

🖲️ VGG Transformer 🖲️

👉VGGT by VGG & #META (#CVPR2025) is a feed-forward neural net. that directly infers all key 3D attributes of a scene within seconds. Code released💙

👉Review https://t.ly/WoWXL
👉Paper https://arxiv.org/pdf/2503.11651
👉Project https://vgg-t.github.io/
👉Code github.com/facebookresearch/vggthttps://t.ly/WoWXL

🦖 DINOv3 is out 🦖

👉#Meta unveils DINOv3! A novel foundation model outperforming the previous SOTAs in computer vision. Code & weights released under DINOv3 License💙

👉Review https://t.ly/-S3ZL
👉Paper https://t.ly/ervOT
👉Project https://lnkd.in/dHFf3esd
👉Repo https://lnkd.in/dPxhDxAq
🤗HF https://lnkd.in/dWGudY2i

🫔ATLAS: SOTA Human Model🫔

👉#META presents ATLAS, a novel high-fidelity body model learned from 600k high-res. scans captured using 240 synchronized cams. Code announced, to be released💙

👉Review https://t.ly/0hHud
👉Paper arxiv.org/pdf/2508.15767
👉Project jindapark.github.io/projects/atlas/
👉Repo TBA

❤️‍🔥PHD: Personalized 3D Humans❤️‍🔥

👉ETH & #Meta unveil PHD, a novel approach for personalized 3D human mesh recovery (HMR) and body fitting that leverages user-specific shape information. Code & models to be released💙

👉Review https://t.ly/IeRhH
👉Paper https://arxiv.org/pdf/2508.21257
👉Project https://phd-pose.github.io/
👉Repo TBA