Use Case

Drone and AI Equipment Inspection PoC at Tokatsu Techno Plaza

With the cooperation of Tokatsu Techno Plaza, we conducted a PoC for indoor wall inspection using drones. AI analyzes footage captured by the drone to automatically extract and visualize fine cracks that can be easy to miss in visual inspection. Areas at height or locations that normally require scaffolding can be inspected quickly from drone flight footage, supporting labor-saving equipment maintenance and early anomaly detection.

Demo Video

See how AI analyzes drone footage and automatically visualizes cracks.

Detection in the Field

Comparison of original drone footage on the left and AI crack detection results on the right
Left: original footage / Right: AI detection results in orange
Inspection drone flying in a stairwell at Tokatsu Techno Plaza
Drone in the inspection site

HOW IT WORKS

How Crack Detection Works

From wall-surface video recorded by a drone, the system extracts linear features and automatically detects cracks.

Three-step crack detection algorithm: 1 original captured image, 2 linear feature extraction with cracks highlighted in orange, 3 detection and visualization
1. Captured image (original) / 2. Linear feature extraction with cracks highlighted in orange / 3. Detection and visualization

01 Capture

The drone captures wall surfaces in high resolution, including areas at height and narrow spaces.

02 Linear Feature Extraction

Image analysis enhances contrast and extracts the thin, continuous linear structures characteristic of cracks.

03 Detection and Visualization

The system identifies cracks while excluding handrails, steps, joints, and other structures. Detected areas are highlighted and recorded as inspection data.

Key Characteristics

  • Analysis is focused on wall-surface regions to reduce false detections from surrounding structures.
  • The system detects cracks stably across continuous video frames.
  • Captured data can be accumulated as inspection records and used to track changes over time.

Summary

In this PoC, we validated automated crack detection for wall inspection by combining drone-based image capture with AI image analysis. The approach complements manual visual inspection and confirmed the potential to inspect elevated or wide-area facilities in a shorter time.

a.s.ist supports Physical AI projects from PoC through implementation by combining drones, cameras, edge AI, and analysis algorithms according to each site's equipment, data, and operating conditions.

Contact Us

Contact us if you are considering a PoC for equipment inspection, drone utilization, or image analysis AI.

  • Detection model design matched to inspection targets and capture conditions
  • PoC validation using drone footage and camera images
  • Visualization and reporting implementation for field operations
Contact Us