WORKFLOWS / KNOWLEDGE & DESIGN
Facade Capture Planning and Quality | Fewer Reshoots
Sunlight, reflections, positions, capture conditions, and matching images to drawings. Build an acquisition plan that won't leave the person interpreting it later stuck.
Read this page in order
Decide what you want to observe before you think about images
Do you want to see the visible appearance of cracks, or locate areas with a temperature difference using infrared? Different purposes need different fields of view and conditions. MLIT's (Japan's Ministry of Land, Infrastructure, Transport and Tourism) PLATEAU pilot projects cover planning support that accounts for sunlight and reflections. Don't assume that mounting a sensor gives you the same result every time.
Source: MLIT PLATEAU | Support for facade inspectionSource: MLIT | Periodic reporting system and facade inspectionKeep a record of the capture plan
| Item | What to record | Problem when it's missing |
|---|---|---|
| Position mapping | Building ID, face, zone, orientation | The location of an anomaly can't be pinned down |
| Conditions | Time, sunlight, weather, distance and angle | You can't tell deterioration from a change in conditions |
| Equipment | Aircraft, sensor, settings, required calibration data | Comparability can't be assessed |
| Coverage | Planned zones, captured zones, obstructions | Gaps in coverage stay hidden |
| Originals | Mapping between acquired files and processed files | You can't get back to the unprocessed evidence |
Thresholds depend on the equipment, materials, method and site conditions. Rather than a general-purpose site handing out one flight distance or one temperature difference as the right answer, a better design records the requirements of the method you apply and who verified it.
Quality checks that close out on site
- Check against a list that every planned face and zone is covered.
- Treat blur, blown highlights, obstructions and unknown positions as separate issues.
- Pair visible-light and infrared images so the same spot can be found in both.
- Re-check the weather and flight conditions to see whether additional capture is possible.
- Hand spots that can't be re-captured to the inspection lead, with the reason.
AI can help pull out candidates for missing or blurred images. But don't adopt an image unconditionally just because the AI rated it "good"; add a step that confirms it can actually be used for the intended interpretation.
Example upload design
Receive original
→ record hash, size and format
→ extract capture time and equipment metadata
→ confirm mapping to a zone
→ save quality-check results
→ confirm as ready for interpretation
Same file sent again: detect the duplicate and check the linkage
Position mapping unknown: hold in a review queue instead of sending to analysisThis is a spec example for a future app. This reference site has no feature for uploading images. When you implement it, define the permissions and deletion rules for originals at the same time.
DEFINITION OF DONE
When you build it, check these.
- The capture status of every zone is visible, and the reason for any uncaptured zone is recorded.
- You can get from a processed image back to its original.
- Duplicate uploads don't double up findings.
Design checklist items for spec reviews and functional testing.
Check the sources
Explanations of regulations and definitions refer to the material below. Screens, data and the definition of done are our own design proposals.
- MLIT PLATEAU | Support for facade inspectionPilot project on inspection planning that accounts for sunlight, reflections and more
- MLIT | Periodic reporting system and facade inspectionPositioning of infrared inspection and the guidelines
Sources checked on September 30, 2026. For your specific case, please confirm against the latest official information.