Everything runs locally on your machine. No cloud services, no uploads, no third parties.
Two specialised neural networks — one optimised for faces, one for license plates — trained on real street-level and equirectangular imagery. No off-the-shelf photo models. No adaptation hacks. Built from the ground up for 360° panoramas.
Inspect every face and plate detection at full resolution. Add missed regions, remove false positives, fine-tune results before export. Complete human oversight without sacrificing the speed of automation.
Add custom watermarks such as © YourCompany to every processed image. Full control over font, size, rows, columns, rotation, opacity, and colour. Perfect for branding deliverables before client handoff.
Create permanent masks to hide static areas — recording vehicle roofs, equipment mount points, branding. Save different mask profiles for different vehicles. Customise mask colours per profile. No per-image rework.
Every pixel of metadata survives processing — GPS coordinates, camera make/model, lens data, heading, timestamps. Original EXIF is copied byte-for-byte. No stripped headers. No broken workflows for Google Street View, Mapillary, or client deliverables.
Command-line interface for scripted, headless automation. JSON output mode for integration into CI/CD pipelines, server environments, and batch processing workflows. No GUI required — process thousands of panoramas programmatically.
NVIDIA CUDA acceleration processes a full Mosaic 51 12K panorama in ~9 seconds. Insta360 X4 in ~4.3s. Automatic CPU fallback if no GPU is available — zero configuration required. Scales efficiently from 4GB to 24GB VRAM.
Zero telemetry. Zero network calls. Zero cloud processing. Your raw panoramas never leave your drive — unlike cloud-based alternatives that require uploading sensitive street-level data to third-party servers. Original files are never modified.
| Feature | Free Trial | License |
|---|---|---|
| Face detection & blurring | ? | ? |
| License plate detection & blurring | ? | ? |
| NVIDIA CUDA acceleration | ? | ? |
| CPU fallback | ? | ? |
| Mask Editor | ? | ? |
| Watermarks | ? | ? |
| Manual review & editing | ? | ? |
| EXIF & metadata preservation | ? | ? |
| Multi-language support (EN / DE / FR) | ? | ? |
| Audit log | ? | ? |
| Batch processing | ? | ? |
| Free updates | — | ? |
| Priority support | ? | ? |
During the trial period, watermarks automatically include "© PanoHopper" in addition to your custom text. Trial masks also include a PanoHopper logo overlay.
Select your input folder containing 360° panoramas in JPEG format.
Click Start. GPU acceleration processes panoramas in seconds. Review detections and make manual edits if needed.
Anonymized images are saved to your output folder with EXIF, GPS, and camera metadata preserved. Original files remain untouched.
| Camera / Format | Resolution | GPU Time | CPU Time |
|---|---|---|---|
| Mosaic 51 | 12,288 × 6,144 px | ~9 s | ~26 s |
| Insta360 X4 | 8,192 × 4,096 px | ~4.3 s | ~15 s |
| GoPro MAX | 5,760 × 2,880 px | ~2.3 s | ~9 s |
Benchmarks measured on an NVIDIA RTX 3090 GPU and AMD Ryzen 9 7950X3D CPU.
Performance scales efficiently across GPUs from 4 GB to 24 GB VRAM. Actual performance may vary depending on hardware configuration.
Developed and supported in Langen. Built with German engineering standards.
No data collection, no telemetry, and no cloud processing. Your images stay on your own storage.
All processing happens locally on your hardware. Your data remains private at all times.
Every processing run generates a detailed log file including system information, settings, per-image results, and performance metrics.
Start your 7-day free trial today. No credit card required. All features unlocked — try before you buy.
A step-by-step workflow for detecting and blurring faces across equirectangular panoramas at scale.
Read guide ?How multi-scale AI detection catches plates at the seams of a 360° image without distorting your output.
Read guide ?What the regulation requires when you publish images of public spaces, and how offline anonymisation helps.
Read guide ?