Уже доступно! Исследование Telegram 2025 — ключевые инсайты года 
AI with Papers - Artificial Intelligence & Deep Learning
Открыть в Telegram
All the AI with papers. Every day fresh updates about #DeepLearning #MachineLearning #LLM & #ComputerVision Curated by Alessandro Ferrari | https://www.linkedin.com/in/visionarynet/ #AI #chatGPT
Больше📈 Аналитический обзор Telegram-канала AI with Papers - Artificial Intelligence & Deep Learning
Канал AI with Papers - Artificial Intelligence & Deep Learning (@ai_deeplearning) языкового сегмента Английский является активным участником. Сейчас сообщество объединяет 17 168 подписчиков, занимая 7 718 место в категории Технологии и приложения и 2 234 место в регионе Малайзия.
📊 Показатели аудитории и динамика
С момента создания невідомо проект демонстрирует стремительный рост, собрав аудиторию из 17 168 подписчиков.
Согласно последним данным от 20 июня, 2026, канал показывает стабильную активность. За последние 30 дней изменение числа участников составило -169, а за последние 24 часа — 0, при этом общий охват остаётся высоким.
- Статус верификации: Не верифицирован
- Уровень вовлечённости (ER): Средний показатель вовлечённости аудитории составляет 22.86%. В первые 24 часа после публикации контент обычно набирает N/A% реакций от общего числа подписчиков.
- Охват публикаций: В среднем каждый пост получает 3 926 просмотров. В течение первых суток публикация набирает 0 просмотров.
- Реакции и взаимодействия: Аудитория активно поддерживает контент: среднее количество реакций на один пост — 26.
- Тематические интересы: Контент сосредоточен на ключевых темах, таких как framework, object, dataset, tba, depth.
📝 Описание и контентная политика
Автор описывает ресурс как площадку для выражения субъективного мнения:
“All the AI with papers. Every day fresh updates about #DeepLearning #MachineLearning #LLM & #ComputerVision
Curated by Alessandro Ferrari | https://www.linkedin.com/in/visionarynet/
#AI #chatGPT”
Благодаря высокой частоте обновлений (последние данные получены 21 июня, 2026) канал поддерживает актуальность и высокий уровень охвата публикаций. Аналитика показывает, что аудитория активно взаимодействует с контентом, что делает его важной точкой влияния в категории Технологии и приложения.
17 168
Подписчики
Нет данных24 часа
-357 дней
-16930 день
Архив постов
🦗Character Mixing Generation🦗
👉MBZUAI unveils the first ever video-gen system able to preserve character ID, behavior & original style while generating plausible interactions between characters that have never coexisted - from cartoons (We Bare Bears, Tom & Jerry) to realistic humans (Mr. Bean, Young Sheldon)
👉Review https://t.ly/tN84a
👉Paper https://lnkd.in/dhKMwukv
👉Project https://lnkd.in/dBkJs48h
👉Repo https://lnkd.in/dw_uzgAk
🐠ITTO: Protocol for Dynamic Tracking🐠
👉ITTO by Caltech is a novel long-range tracking benchmark suite for evaluating and diagnosing tracking methods on complex and long-range motions. Repo under CC BY-NC 4.0💙
👉Review https://t.ly/tN84a
👉Paper https://arxiv.org/pdf/2510.19819
👉Project https://glab-caltech.github.io/ITTO/
👉Repo https://github.com/ilonadem/itto
🏜️Omni Driving Navigation Models🏜️
👉OmniNWM is a unified panoramic navigation world model that advances autonomous driving by jointly generating multi-modal states (RGB, semantics, depth, 3D occupancy), enabling precise action control & facilitating closed-loop evaluation through occupancy-based dense rewards. Repo under Apache 2.0💙
👉Review https://t.ly/ktXvz
👉Paper https://lnkd.in/eFKSZnrc
👉Project https://lnkd.in/eSDfccv8
👉Repo https://lnkd.in/efCSvjtp
Repo (pretty empty) now online: https://github.com/OatmealLiu/UrbanVerse
🔥 SAM 2++: Track Anything 🔥
👉SAM 2++ is a novel unified model towards tracking at any granularity, including masks, boxes, and points. Impressive results but no code announced yet 😢
👉Review https://t.ly/I392_
👉Paper arxiv.org/pdf/2510.18822
👉Project tracking-any-granularity.github.io/
👉Repo :(
+2
🌵All-in-One Dense Keypoints🌵
👉DeepDetect is a novel all-in-one, dense keypoints detector that unifies the strengths of SIFT, ORB, BRISK, FAST, AGAST, Harris, Shi-Tomasi, Canny & Sobel into a neural net. DAMN ROMANTIC. Repo under MIT💙
👉Review https://t.ly/VKGct
👉Paper https://arxiv.org/pdf/2510.17422
👉Repo https://github.com/saktx/DeepDetect
🦄 City-Tour -> Simulation 🦄
👉UrbanVerse is a novel system to convert real-world urban scenes from city-tour videos into physics-aware, interactive simulation environments, enabling scalable robot learning in urban spaces with real-world generalization. Repo & Data announced 💙
👉Review https://t.ly/UvXNS
👉Paper https://arxiv.org/pdf/2510.15018
👉Project https://urbanverseproject.github.io/
👉Repo TBA
🫙Universal Feature Up-Sampling🫙
👉AnyUp is a novel method for feature up-sampling that can be applied to ANY vision feature at ANY resolution, without encoder-specific training: inference-time feature-agnostic up-sampling architecture to improve up-sampling quality. Repo under CC-4.0💙
👉Review https://t.ly/HvEw9
👉Paper https://arxiv.org/pdf/2510.12764
👉Project https://wimmerth.github.io/anyup/
👉Repo https://github.com/wimmerth/anyup
🫧🫧 Detect Anything via MLLM 🫧🫧
👉Rex-Omni is a 3B-multimodal model that unifies visual perception tasks, including object detection, OCR, pointing, key-pointing & visual prompting into a single next point prediction framework. Impressive results. Full repo under IDEA License 1.0💙
👉Review https://t.ly/DCTk_
👉Paper https://lnkd.in/d4VDD-9j
👉Project https://lnkd.in/d6unEyvq
👉Repo https://lnkd.in/dkYJFe-x
↗️ TrackVLA++ Visual Tracking↘️
👉TrackVLA++ is a novel Vision-Language-Action model that incorporates spatial reasoning and target identification memory, enabling SOTA performance in both long-horizon and highly crowded tracking scenarios. Model announced💙
👉Review https://t.ly/ruYzc
👉Paper https://arxiv.org/pdf/2510.07134
👉Project pku-epic.github.io/TrackVLA-plus-plus-Web/
👉Repo TBA
💄Pixel-Perfect Depth (SOTA)💄
👉Pixel-Perfect Depth is a mono-depth estimation model with pixel-space diffusion transformers. New SOTA. Repo under Apache 2.0💙
👉Review https://t.ly/75PGo
👉Paper https://lnkd.in/d8wxFpyY
👉Project https://lnkd.in/dV5HhsqH
👉Repo https://lnkd.in/d9JKFBJq
👉Demo https://lnkd.in/d3wBkKJ9
🎺Visual Grounding RVOS🎺
👉ReferDINO is a strong RVOS model that inherits region-level vision-language alignment from foundational visual grounding models, and is further endowed with pixel-level dense perception & cross-modal spatio-temporal reasoning. Code, Demo & checkpoints released💙
👉Review https://t.ly/rOdkP
👉Paper https://lnkd.in/efuAFQdE
👉Project https://lnkd.in/dK3wMZqv
👉Repo https://lnkd.in/d3i2PsNF
🎺Visual Grounding RVOS🎺
👉ReferDINO is a strong RVOS model that inherits region-level vision-language alignment from foundational visual grounding models, and is further endowed with pixel-level dense perception & cross-modal spatio-temporal reasoning. Code, Demo & checkpoints released💙
👉Review https://t.ly/rOdkP
👉Paper https://lnkd.in/efuAFQdE
👉Project https://lnkd.in/dK3wMZqv
👉Repo https://lnkd.in/d3i2PsNF
👉 A proof I'm not a bot...
My (short) interview to one of the biggest Italian media: AI in 2016, HPC / Quantum and how I created my startup: https://www.linkedin.com/posts/visionarynet_ai-itw25-ai-activity-7381215486115643392-t7an
Thanks for the support (and of course a new paper coming in a few hours)
🎷🎷 Clink! Chop! Thud! 🎷🎷
👉Sounding Object Detection: while an environment may contain many objects, only a few are directly involved in producing sound during an interaction. This model detects the sounding object given a video of an object interaction. Code/Data announced💙
👉Review https://t.ly/VK_1h
👉Paper https://lnkd.in/depNjVXm
👉Project https://lnkd.in/dF63EZFG
👉Repo TBA
🔩Code-Centric Agentic Education🔩
👉Show Lab unveils Code2Video: agentic, code-centric framework that generates HQ educational videos from knowledge points. Unlike pixel-based text-to-video models, this approach leverages executable Manim code to ensure clarity, coherence & reproducibility. Repo under MIT💙
👉Review https://t.ly/Fv4LJ
👉Paper https://arxiv.org/pdf/2510.01174
👉Repo https://github.com/showlab/Code2Video/
👉Project https://showlab.github.io/Code2Video/
👩🦱Physical-Hair Diffusion👩🦱
👉CONTROLHAIR is novel hybrid framework that integrates a physics simulator with conditional video diffusion to enable controllable dynamic hair rendering. Repo announced💙
👉Review https://t.ly/78LHr
👉Paper https://lnkd.in/epm-A9Fq
👉Project https://lnkd.in/evsjz298
👉Repo TBA
👔 Universal Image Restoration 👔
👉LucidFlux by HKUSTGZ is the universal image restoration framework built on a large-scale diffusion transformer that delivers photorealistic restorations of real-world low-quality (LQ) images, outperforming SOTA diffusion-based models across diverse degradations. Repo under custom Non-Commercial License💙
👉Review https://t.ly/Z5cA3
👉Paper https://arxiv.org/pdf/2509.22414
👉Project https://w2genai-lab.github.io/LucidFlux/
👉Repo https://github.com/W2GenAI-Lab/LucidFlux
🤖 Real-time Interactive Long Video 🤖
👉LONGLIVE by #Nvidia is a frame-level autoregressive framework for real-time & interactive long video generation. LONGLIVE accepts sequential user prompts and generates corresponding videos in real time. Repo under non-commercial license💙
👉Review https://t.ly/jJkdY
👉Paper arxiv.org/pdf/2509.22622
👉Project nvlabs.github.io/LongLive/
👉Repo github.com/NVlabs/LongLive
🤗huggingface.co/Efficient-Large-Model/LongLive-1.3B
🔥SOTA Detection w/ DINOv3🔥
👉DEIMv2 is the evolution of DEIM framework while leveraging DINOv3. Various model sizes, from an ultra-light version up to S, M, L, & X for a wide range of scenarios. Across these variants, DEIMv2 achieves SOTA. Repo Apache2.0💙
👉Review https://t.ly/P7jEH
👉Paper arxiv.org/pdf/2509.20787
👉Repo github.com/Intellindust-AI-Lab/DEIMv2
👉Project intellindust-ai-lab.github.io/projects/DEIMv2
