This is a POC application to detect the analog meter reading in an image. Visual inspection is necessary to obtain readings from analog gauges. In many industries, this means a human operator must travel to the gauge’s location, read its current value, and log that value to enable the data to be used elsewhere. The ability to read an analog gauge using computer vision allows for the integration of readings into an automated system, which stands to deliver significant advantages, including:

• Reducing the need for a maintenance crew to travel to remote locations
• Providing near real-time, continuous access to gauge levels
• Allowing for digital monitoring of gauge readings and automatic alerts when readings go out of tolerance

In this example, we leveraged deep learning techniques to automatically detect and monitor physical analog gauges to produce streams of digital readouts and events.

The URL for gauge reader is – https://gauge.sakhaglobal.com:7000/

If you want to upload an image – Click “Gauge Image Detection”

• You can upload a Gauge image (preferably .JPG images) of your choice.
• You will be redirected to the next page, where the calibrated gauge image is displayed. Calibrated image contains the angles at which gauge readings are marked.
• Observing the calibration fill the minimum & maximum gauge value, and minimum & maximum angles of gauge.
• The algorithm will calculate and gives the final reading.

If you want to upload a gauge video – Click “Gauge Video Detection”

• Upload a gauge video (preferably .mp4).
• You will be redirected to the next page, where the calibrated gauge image is displayed. Calibrated image contains the angles at which gauge readings are marked.
• Observing the calibration fill the minimum & maximum gauge value, and minimum & maximum angles of gauge.
• The uploaded video is played, annotated with respective gauge readings.