How to Extract Text from Photos
Photos of signs, documents, whiteboards, and screens contain valuable text. With the right technique, you can extract that text in seconds using OCR — no typing required.
Getting a Good Photo for OCR
The quality of OCR output is directly determined by the quality of the input photo. Camera photos are inherently noisier than flatbed scans, but following these tips can dramatically improve accuracy:
- Lighting is everything. Even, diffused lighting produces the best results. Avoid direct flash (creates hotspots) and strong shadows (reduce contrast).
- Keep the camera parallel. Shoot straight-on to avoid perspective distortion. Angled photos cause text to shrink and skew, reducing accuracy.
- Fill the frame with text. The more pixels per character, the better the OCR. Zoom in or move closer rather than cropping later.
- Hold steady. Blur from camera shake is the #1 killer of photo OCR accuracy. Use both hands, brace against a surface, or use a timer.
- Use the highest resolution. Modern phones capture 12-48 MP. Make sure your camera app isn't saving in a reduced resolution mode.
Photo OCR Accuracy by Source
| Photo Source | Typical Accuracy | Key Challenge |
|---|---|---|
| Document flat on desk | 92–96% | Slight curvature |
| Whiteboard photo | 85–92% | Glare and marker thickness |
| Street sign / poster | 80–90% | Distance and angle |
| Phone screen photo | 88–94% | Screen glare and moire patterns |
| Book page photo | 82–90% | Page curvature near spine |
When to Use Photo OCR vs Scanning
Photo OCR is ideal for quick, informal text capture — signs, whiteboards, quick document snapshots. For important documents where accuracy is critical, a flatbed scanner or document scanning app produces better results. The tradeoff is speed (photos are instant) vs accuracy (scans are 5-10% more accurate).
Extract Text from Photos — Free
FastOCR extracts text from photos in seconds. Works in any browser on any device. No registration required.