Testing Google Vision for weather data rescue

As AWS Textract looks promising as a tool for data rescue, it’s worth trying Google Vision in the same way.

I tested Google Vision on sample images from several different documents containing weather observations we need to transcribe. All were somewhat successful, though the level of success varied:

• Sample from the Weatherrescue Ben Nevis project
• Sample from the UK Second Order Stations Books
• Sample from the Argentine Daily Weather Reports
• Sample from the UK Observatories Year Book
• Sample from the UK Daily Weather Reports (1901)
• Sample from the UK Daily Weather Reports (1862)
• Sample from the Indian Daily Weather Reports
• Sample from the US Manual Analysis maps
• Sample from the logbook of USS Farragut
• Sample from the logbook of USS Jeannette

As the systemproduced reasonable results on the Ben Nevis project sample, it was possible to run Google Vision against the OCR-weatherrescue transcription benchmark:

• OCR-weatherrescue benchmark comparison

Results are qualitatively similar to those from AWS Textract.

• Small Print

This document is crown copyright (2019). It is published under the terms of the Open Government Licence.