SEMI-AUTOMATED CRICKET BROADCAST HIGHLIGHT GENERATION
Sahil Ramlogan, Akash Pooransingh, Daniel Joseph Ringis · 2020
This paper provides an improvement to automated cricket highlight identification from full broadcasts. Further refinement is provided through semi- automated user verification. This visual result is synchronized with an automated text extraction process. The visual result was successfully extracted at a rate of 97.5% using the ORB function. With user validation a false detection rate of 0% was realized. The metadata from the commentary was then successfully extracted and combined with the visual results to allow for easy searching of the highlight video using any search field.