• 16 min read
DeepSeek Vision Models
Learn about DeepSeek’s vision-capable models, and how to use DeepSeek-VL2 with Roboflow Supervision to build advanced vision-language applications.
Blog
Timothy M
Latest Posts by Timothy M
• 18 min read
Visual Anomaly Detection
Learn what anomaly detection is, the main methods used in visual anomaly detection, and how to build supervised, and unsupervised anomaly pipelines in Roboflow Workflows for real-time deployment on the edge.
• 21 min read
How to Build Automated Pallet Accounting at End-of-Line with Roboflow
Automate pallet accounting at end-of-line with Roboflow Workflows and RF-DETR. Track every pallet from conveyor arrival to forklift pickup using zone-based monitoring and real-time event logging.
• 14 min read
How to Increase Inference Speed for Computer Vision Models
Struggling with low FPS in your computer vision model? This guide explains how to move from single-digit performance to real-time deployment using smarter preprocessing, Nano-scale models like RF-DETR, GPU acceleration, TensorRT optimization, and Roboflow Inference pipelines for maximum throughput.
• 15 min read
Roboflow Training Graphs Guide
Learn how to read Roboflow training graphs. Understand what each curve means, how the underlying metrics are computed, and how to diagnose common training problems.
• 7 min read
How Do I Monitor Inference Health?
Inference health monitoring is necessary to keep computer vision systems reliable in production. This guide explains key signals like latency, uptime, drift, and confidence trends, and shows how Roboflow helps teams track, diagnose, and improve real-world model performance.
• 10 min read
How Do I Train a Model for Defects I Almost Never See?
Learn how to train and deploy a rare defect detection model with Roboflow.
• 7 min read
Segment Anything
This guide breaks down the SAM lineage - from SAM 1’s 50ms CPU inference to SAM 3’s open-vocabulary "GPT moment" - showing you how to deploy these foundation models into your production workflows today.
• 11 min read
How to Reduce Scrap in Manufacturing with Computer Vision
Scrap isn't inevitable, it is a visibility problem. Most manufacturers discover defects hours after they start. Computer vision changes this by detecting issues in real-time.
• 7 min read
Using Vision-Language Models for Image Understanding
Vision-language models make image understanding fast and accessible. By using simple text prompts, teams can explore datasets, discover labels, spot errors, and speed up annotation early, before training, building stronger foundations for reliable computer vision systems.
• 9 min read
Open-Vocabulary Object Detection Explained
Learn how open-vocabulary object detection actually works, how it differs from promptable segmentation and vision-language models, and when it’s the right tool for real-world vision systems.
• 12 min read
Object Detection vs Vision-Language Models: When Should You Use Each?
Object detection and vision-language models solve different problems. This guide explains when to use each. And how speed, cost, and reliability differ.
• 10 min read
How to Improve Cycle Time with Computer Vision
Learn how computer vision measures cycle time continuously, reveals hidden waiting and handoffs, and turns cycle time from a guessed metric into a live operational signal.
• 7 min read
Spatial Intelligence
As AI systems move beyond screens and into the physical world, understanding three-dimensional space becomes essential. This article explains spatial intelligence: how AI builds world models, reasons about depth and motion, and moves from recognizing images to acting reliably in real environments.
• 6 min read
How to Monitor and Improve AI Models in Production
Learn how to monitor a model in production, and improve its performance.
• 12 min read
Machine Vision Lighting Types
Good lighting is the foundation of any machine-vision system. This guide explains how different lighting types, filters, and deployment choices shape image quality.
• 12 min read
Vision AI Camera Calibration Guide
Learn what camera calibration is, and see how to calibrate a camera step by step with Roboflow.
• 12 min read
Vision Transformer vs. CNN for Object Detection
This guide explains how CNNs and ViTs differ, when to use each, and how to train and compare them directly inside Roboflow so you can choose the best architecture for your detection task.
• 15 min read
Meta SAM 3D: Introduction
Understand what is SAM 3d and learn how to use SAM 3D + Roboflow to generate synthetic data and train a computer vision model.
• 10 min read
What Is Promptable Concept Segmentation (PCS)?
Discover the concept of Promptable Concept Segmentation (PCS), how is it compares with other segmentation tasks such as semantic, instance, and panoptic, and the SAM 3 architecture, data and training infrastructure that makes PCS possible.
• 10 min read
Inference Latency
Learn about inference latency, why it matters, and how to optimize every stage of the pipeline to build reliable, real-time vision systems.
• 13 min read
How Much Training Data Do You Need to Train a Computer Vision Model?
Discover how to find the right amount of training data for your computer vision project. Learn how dataset size, balance, and annotation quality impact model accuracy.
• 30 min read
Best Mobile Object Detection Models
Discover some of the best mobile object detection models, and how to easily build iOS app with Roboflow iOS SDK.
• 7 min read
The Best Data Labeling Tools
Accurate data labeling is essential for training machine learning models. This guide explores top data labeling platforms, key features to consider, and pros and cons so you can choose the right tool for building better AI.
Stay Connected