🚀New dataset on Hugging Face!. FishGrade🐟for real-time, non-invasive grading of fish steaks using RGB+depth and deep learning. Annotations for segmentation + quality classes. By @TeV_FBK, part of #AgileHandEU🇪🇺. 🔗 #AgriTech #AI4Seafood #ComputerVision

RT @FrancisEngelman: Jaime's talk on "Functionality Understanding in 3D Scenes" is now available on YouTube 🎞️🍿: ..

🚀TeV @ IROS 2025.We’re thrilled to announce that 4 papers from our unit will be presented at #IROS2025!. Our work spans language-driven reasoning, navigation, and 6D pose estimation in robotics and AI:. FMF-SLAM.FreeGrasp.dGeDi.GRASPLAT. Congrats to all involved!🙌. #IROS2025

🚨 #Automatica2025 is ending soon — here’s a sneak peek of what’s happening at Booth A06!.Catch @Poiex & @mbortolon97 showing how our #xPerception system powers the @PeitianTech888 arm 🦾.Thanks to Metaup & @FBK_research !. #AI #Robotics #DeepLearning #ComputerVision

🎥Our colleague @mbortolon97 is setting up the camera of our #xPerception system — curious to see it in action?. Join us at #Automatica2025 Booth A06 and see how our vision system lets the @PeitianTech888 arm perceive & interact with the world!. @FBK_research #AI #Robotics #CV

Proud to share the story behind xPerceptio on ilT Quotidiano, as one of the winners of PoC by FBK: Test your Idea!🎙️. From lab to real-world impact—bringing FreeZe closer to deployment!.Thanks to @FBK_research and all who supported us!🙌.#xPerceptio #FreeZe #AI #Robotics #6DPose.

RT @FBK_research: Un’IA che legge la maturazione dei lamponi? .Con #RaspGrade, @TeV_FBK e Sant'Orsola usano la visione artificiale per clas….

We benchmarked 3 zero-shot 6D pose methods on CHIP, revealing significant challenges and room for improvement. 📄Read the paper: 📢Dataset release coming soon!. @FBK_research @IKERLANofficial @AndreuWorld @AIPRISMEU .#6DPoseEstimation #CV #Dataset.

🔍 CHIP highlights:. • 7 visually similar chairs with fine-grained differences. • 77k+ RGB-D images from 3 sensor types. • Ground-truth poses from robot kinematics. • Real challenges: occlusions, distractors, and human operators.

📢 Coming soon: CHIP — a new dataset for 6D pose estimation in industrial settings. It features chairs manipulated by a robotic arm in a real-world environment, pushing the limits beyond typical household object benchmarks. 🧵👇

🎉Great days for the team at #CVPR2025 in Nashville!. We presented multiple papers and had inspiring exchanges with the computer vision community. 📄Check out our work:. @FBK_research

Kudos to the authors: Andrea Caraffa, Davide Boscaini, @Poiex . @FBK_research . #6DPoseEstimation #ComputerVision #ZeroShotLearning #BOPChallenge.

🏆 Best Overall Method @ BOP Challenge 2024. 📊 SOTA on the BOP Benchmark for unseen objects.What's new in V2:. • 8× faster inference. • +5% accuracy (feature-aware scoring). • +8% accuracy (segmentation ensemble).📄 Report:

📌 A new milestone in zero-shot 6D pose estimation. Meet FreeZeV2 — a training-free method for estimating the 6D pose of unseen objects from RGB-D data. No training. No finetuning. No synthetic data. Just geometry + frozen vision foundation models. 🧵👇

🎥“We deliver custom vision-based solutions using traditional & deep learning techniques.”. — @Poiex , Head of TeV (@FBK_research ). In 90 sec, discover our research in:.📌6D pose estimation.📌3D scene understanding.📌Robotic grasping & more. Watch now 👇. #CV #AI #DL #TeV #FBK

What a day!🎉.CVTS2025 brought together cutting-edge research, exciting demos, and a fantastic CV community in Trento. 📸Pics are online:Thanks to everyone who joined, and special thanks to @AgileHand_ , @fondazione_fair ,and @FBK_research for the support!

RT @FBK_research: CVTS 2025: a Trento il meglio della computer vision.Quasi 100 partecipanti a FBK per la seconda edizione del symposium or….

🤖Just a few days to go until CVTS 2025 and things are starting to warm up!.Here’s a sneak peek at the FreeGrasp demo in action, as the robot arm gets ready to show off its skills🔧.📍@FBK_research May 29 .🎟️Register now:.🌐Info:

🚨Just 1 week to #CVTS2025!. 📅May29 |📍Trento. 🎤Keynote by @lambertoballan on embodied visual navigation. 🔹 17 CVPR 2025 presentations.🔹 Posters, talks & live demos. 🌐Updated program: 🙏Thanks to @AgileHand_ , @fondazione_fair & @FBK_research!. #CV

🎓Great energy at #PhDDay2025 @FBK_research!.🚀Huge kudos to @amirhamza848, for presenting dGedi: a lightweight 3D point cloud encoder pushing zero-shot 6D pose estimation closer to real-time!🔬🦾.📌170× faster.📌Robust to occlusion.📌Ideal for robotics.🔗