In this thesis a novel audiovisual feature-based scheme is proposed for the automatic summarization of sports-video content The scope of operability of the scheme is designed to encompass the wide variety o f sports genres that come under the description ‘field-sports’. Given the assumption that, in terms of conveying the narrative of a field-sports-video, score-update events constitute the most significant moments, it is proposed that their detection should thus yield a favourable summarisation solution. To this end, a generic methodology is proposed for the automatic identification of score-update events in field-sports-video content. The scheme is based on the development of robust extractors for a set of critical features, which are shown to reliably indicate their locations. The evidence gathered by the feature extractors is combined and analysed using a Support Vector Machine (SVM), which performs the event detection process. An SVM is chosen on the basis that its underlying technology represents an implementation of the latest generation of machine learning algorithms, based on the recent advances in statistical learning. Effectively, an SVM offers a solution to optimising the classification performance of a decision hypothesis, inferred from a given set of training data. Via a learning phase that utilizes a 90-hour field-sports-video trainmg-corpus, the SVM infers a score-update event model by observing patterns in the extracted feature evidence. Using a similar but distinct 90-hour evaluation corpus, the effectiveness of this model is then tested genencally across multiple genres of fieldsports- video including soccer, rugby, field hockey, hurling, and Gaelic football. The results suggest that in terms o f the summarization task, both high event retrieval and content rejection statistics are achievable.