appreciate it 🫡 custom yolo bytetrack ocr, all on a single moving camera. no roboflow. your (x,y,z) setup is the dream. we should link.

🚗⚡ What if a regular CCTV camera could measure vehicle speed using AI? This project combines YOLO, ByteTrack, and Computer Vision to detect vehicles, track them in real time, and estimate their speed automatically. 🤯 ✅ Vehicle Detection ✅ Real-Time Tracking ✅ Speed Estimation ✅ Motion Trails ✅ Smart Traffic Analytics The best part? It’s built with open-source tools and runs on standard video footage—no expensive speed camera hardware required. 💬 Comment “SPEED” and I’ll send the project details. 📌 Save this reel for your next AI project. 🔄 Share with a friend who loves AI and Python. #AI #ArtificialIntelligence #ComputerVision #YOLO #Python MachineLearning DeepLearning OpenSource Coding Programmer SoftwareEngineer DataScience Robotics Automation SmartCity VehicleTracking SpeedDetection Tech Engineering AIProjects PythonProjects TechReels FutureTech Developer Innovation

أغلب نماذج الرؤية الحاسوبية (Computer Vision) بتعطيك مجرد مربعات تحديد (Boxes) وأقنعة (Masks)، لكن تطبيقك الفعلي في النهاية بيحتاج عناصر وعمليات أعمق بكثير: إطارات فيديو، معرّفات تتبع (Tracking IDs)، تصنيفات (Labels)، مناطق مخصصة (Zones)، عمليات عدّ، وتقييم شامل للأداء. هنا بيبرز دور أداة Supervision (من تطوير Roboflow). الأداة ببساطة بتوحّد مخرجات أشهر منصات ونماذج الرؤية الحاسوبية مثل Ultralytics، Transformers، MMDetection، و Inference، وتجمعها تحت مظلة واجهة برمجية مشتركة وموحدة وهي (sv.Detections API). وفوق هذا، بتضيف لك أدوات جاهزة ومتقدمة مثل الـ Annotators للتوضيح البصري، وخوارزمية التتبع الشهيرة ByteTrack، وأدوات دقيقة لإدارة النطاقات والـ Zones. الخلاصة المريحة لمهندسي الأنظمة التقنية: 📉 تقليل كود الربط المؤقت والترقيعي (Less demo glue) اللي بتكتبه لمجرد العرض. ⚙️ بناء خط معالجة رؤية حاسوبية (CV Pipeline) مستقر، صلب، وقابل للتكرار والاعتماد عليه في الإنتاج الفعلي. المشروع مفتوح المصدر بالكامل برخصة MIT وحصد أكثر من 43,000 نجمة على GitHub نتيجة كفاءته. #ComputerVision #Roboflow #AIInfra #MENAAI #SaudiAI #JordanTech #ArabAIEra #ATIEHTECH

Cada vez que construyes algo con visión por computadora escribes el mismo código. Bounding boxes. Seguimiento de objetos. Zonas de detección. Métricas de rendimiento. x.com/skalskip92/status/2063… Todo desde cero. Cada proyecto. Siempre. Roboflow lleva años construyendo la librería que elimina ese problema. Se llama Supervision y tiene 42.1k estrellas en GitHub. 1 millón de descargas al mes en PyPI. 6.5k proyectos open source construidos encima. Activo hace 2 horas. Lo que hace en una línea: Conectas cualquier modelo de visión y Supervision se encarga del resto. ✅ Objeto unificado sv.Detections compatible con YOLO, SAM, Transformers, Detectron2, MMDetection y más ✅ Anota imágenes y video con bounding boxes, máscaras y etiquetas en 3 líneas de código ✅ Seguimiento de objetos entre frames con IDs persistentes via ByteTrack y BoTSORT ✅ Cuenta y filtra detecciones dentro de zonas poligonales personalizadas ✅ Estima velocidad de objetos en movimiento con transformación de perspectiva ✅ Carga y convierte datasets entre YOLO, COCO y Pascal VOC ✅ Benchmarks con mAP y matrices de confusión listos para usar ✅ Compatible con Claude Code via AGENTS(.)md integrado ✅ 36 releases. v0.28.0. MIT. El caso de uso que más me ha flipado: El propio autor lo usó para construir un sistema de IA que sigue en tiempo real a los jugadores de baloncesto en un partido. Detección, seguimiento, zonas de cancha y métricas de velocidad. Todo con Supervision y menos de 100 líneas de Python. el enlace 👇

That's not a vanity metric. Here's what it actually means: 6,500 open-source CV projects are built on top of it right now. Basketball AI tracking. Traffic monitoring. Retail analytics. Medical imaging. Drone detection. All running on the same library. For those who don't know — supervision is Roboflow's open-source toolkit for computer vision post-processing. Not a model. Not a dataset. The layer between your model and your application. Bounding boxes, segmentation masks, tracking, annotation, filtering — all in clean, composable Python. import supervision as sv annotator = sv.BoxAnnotator() detections = sv.Detections.from_ultralytics(results) frame = annotator.annotate(frame, detections) That's literally it. Why it blew up: → Works with ANY model — YOLO, SAM, DINO, Grounding DINO, RF-DETR → Tracker-agnostic (ByteTrack, SORT, BoTSORT) → Zero boilerplate for video inference loops → Built-in dataset tools (COCO, YOLO, VOC formats) → Active maintainers, weekly releases The numbers: ⭐ 40,000 stars 📦 6,500 dependent projects 🔄 Used in production across sports, security, agriculture, healthcare From 0 to one of the most-starred CV libraries on GitHub — in under 2 years. If you're still writing your own draw_bounding_box() function in 2025 — this is your sign to stop. → github.com/roboflow/supervis…

YOLOv8x detection ByteTrack tracking (Roboflow Inference & Supervision)

And camera angle What I actually want this to become: proper DeepSORT or ByteTrack so vehicles don’t lose their ID when they cross paths. Real lane detection so speed limits can be applied per lane.

Vehicle counting using @ultralytics YOLO26 🚀 Here, I trained the YOLO26 model on a custom dataset and later used an object tracker (Bytetrack) Ultralytics Solutions to count the objects in the video file. #vehicles #tracking #MachineLearning

Nice build. The custom dataset plus ByteTrack is what makes this feel usable, not just a model demo.

Packet counting using @ultralytics YOLO26 🚀 Here, I trained the YOLO26 model on a custom dataset and later used an object tracker (Bytetrack) Ultralytics Solutions to count the objects in the video file. #packets #counting #MachineLearning

@roboflow keypoint model detects 12 court corners, RANSAC homography maps pixels to a 20×44ft court, ByteTrack keeps persistent player IDs, and TrackNet finds a 5-10px ball across consecutive frames.

Vehicle tracking using @ultralytics YOLO26 👀 In this demo, I trained the YOLO26 model on a custom dataset and later used an object tracker (Bytetrack) to track the objects in the video file. Note: I also performed testing with Botsort, which is also a good tracker, but it's slow in comparision to Bytetrack. #Tracking #ComputerVision #MachineLearning

pip install trackers Ya está disponible trackers v2.1.0; esta versión añade soporte para ByteTrack, un algoritmo rápido de seguimiento por detección enfocado en mantener identidades estables bajo oclusión. Enlace: github.com/roboflow/trackers

We should be nervous about how much power the AI models have in discovery. I’m doing some car speed tracking using a camera on our property and Opus chose an app stack (Python:3.12-slim base image, UV, FastAPI, uvicorn, ) ML stack (Yolo11s, Bytetrack, ONNX) cloud architecture (Cloudflare R2->queue->worker->container->D1). In the before times, I would have spent a whole day researching these pieces, testing them, talking to friends, and building a plan. Now? I gave it a nudge about the cloud arch because I had an opinion, but otherwise, turned “auto mode on” and let it rip. Took a few hours to build. I never opened a browser let alone a search engine. This will start to apply to every field: education, health, politics, history...

We introduce Multi-Resolution Rescored ByteTrack "MR2-ByteTrack: CNN and Transformer-based Video Object Detection for AI-augmented Embedded Vision Sensor Nodes" which reduces computational cost by alternating between full- and low-resolution inference. See arxiv.org/pdf/2605.15423

Built a hybrid AI agent for surveillance: YOLO11 ByteTrack → tracking SQLite → deterministic queries LLM → contextual reasoning “How many cars?” → SQL (exact) “Anything suspicious?” → LLM Same interface. Two execution paths.

Hmmm, I am just using the Ultralytics package, which already supports Bytetrack, so not sure if anything additional I can help with 😀

Football players detection tracking using @ultralytics YOLO26 ⚽️ It took me 30 minutes to train the YOLO26 model on a custom dataset using Ultralytics Platform. Later, I used the object tracker (Bytetrack) for tracking the players; tracking could be improved further, and I am working on it this week. More info 👇 #soccer #football #MachineLearning

Technically, @flypraxis combines YOLO detection with ByteTrack, using temporal buffering and low-confidence association to maintain tracking continuity during short-term occlusions or signal loss scenarios.