Guides & references
Practical guides and technical references for working with data, images, documents, geospatial formats, and developer tools.
Computer Vision
Computer Vision Annotation Formats: Bounding Boxes, YOLO, COCO & Pascal VOC Explained
What YOLO, COCO, Pascal VOC, and VIA CSV annotations actually contain, how their coordinate systems differ, and where conversion and verification fit in a real object detection workflow.
How to Convert Between YOLO, COCO, and Pascal VOC Annotations
Step-by-step conversion between YOLO, COCO, and Pascal VOC annotation formats — including what gets lost in a lossy conversion and how to check the result.
How to Visually Verify Object Detection Annotations Before Training
Common annotation errors that silently hurt model training, and how to spot-check a dataset visually before you trust it.
How to Split a YOLO Dataset Into Train/Val/Test Without Leakage
Why a naive random split leaks data between train and test through near-duplicate images and class imbalance, and how to split a YOLO dataset with a reproducible seed and a leakage check.
How to Remap, Merge, and Clean Up Classes in a YOLO Dataset
Rename, merge, or remove classes across an entire YOLO dataset in one pass, with an impact preview before anything is rewritten and a corrected copy exported rather than touching the original.
YOLO vs COCO vs Pascal VOC: Annotation Format Comparison
A quick side-by-side of file structure, coordinate systems, and tooling for YOLO, COCO, and Pascal VOC annotations.
Geospatial
Geospatial File Formats: GeoJSON, KML, GPX, CSV, WKT, TopoJSON & Shapefile Explained
What GeoJSON, KML, GPX, CSV, WKT, TopoJSON, and Shapefile actually contain, the coordinate-order bug that trips up almost everyone, and where conversion and verification fit in a real geospatial workflow.
How to Convert Between GeoJSON, KML, GPX, and Other Geospatial Formats
Step-by-step conversion between GeoJSON, KML, GPX, CSV, WKT, TopoJSON, and Shapefile — what gets left out converting to GPX or CSV, and how Shapefile reprojection works.
How to Spot Common Geospatial Data Errors Before You Trust a Dataset
The coordinate-order swap, projection, and precision errors that produce valid-looking but wrong geospatial files — and how to catch them visually before they feed into anything downstream.
How to Reproject a Shapefile Without a Desktop GIS
Fix a Shapefile whose coordinates are in the wrong place — pick a UTM zone or paste a CRS definition, reproject to WGS84, and repair invalid geometry, entirely in the browser.
How to Verify a Geospatial Conversion Actually Worked
Why "conversion complete" isn't the same as "conversion correct" — how to see exactly what a format conversion changed, and how to compare any two datasets, not just a conversion pair.
How to Clean Up a GPX Track Before Trusting It
Duplicate GPS fixes, backward timestamps, and impossible speed jumps that a map view won't show you — and which ones are actually safe to fix automatically.
GeoJSON vs KML vs GPX vs Shapefile: Format Comparison
A quick side-by-side of file structure, coordinate systems, and typical use for the four most commonly compared geospatial formats.
Visualization
Choosing the Right Chart Type for Your Data
What bar, line, pie, histogram, scatter, box plot, and heatmap charts each actually answer — and which of Tuliq's chart tools fits which question.
How to Spot Outliers and Correlations in Your Data
A workflow for exploring a dataset with many numeric columns — start broad with a correlation heatmap, zoom into a specific pair with a scatter plot, then check for outliers with a box plot.
Mermaid vs Graphviz: Which Diagram Tool Should You Use?
The real difference between Mermaid's diagram-type-aware syntax and Graphviz's general-purpose DOT layout engine, and which one fits a given diagram.
Local AI
How Local AI Works in Your Browser: On-Device Image Tools Explained
What Tuliq's 🔵 Local AI badge actually means — a real model, downloaded once, running entirely on your device — and how that's different from uploading to a cloud AI service.
How to Use AI Object Detection to Speed Up Annotation
Use the on-device Object Detector to draft bounding boxes automatically, then correct and export them into the same YOLO/COCO/Pascal VOC/VIA CSV pipeline as any other annotation source.