Annotation Techniques for Judo Combat Phase Classification from Tournament Footage
This is an incremental domain-specific solution for automating judo match analysis with limited labeled data.
This paper tackles the problem of automating annotation and summarization of judo matches from tournament footage by developing a semi-supervised approach to classify combat phases, achieving F1 scores of 0.66, 0.78, and 0.87 on a test set of 19 clips.
This paper presents a semi-supervised approach to extracting and analyzing combat phases in judo tournaments using live-streamed footage. The objective is to automate the annotation and summarization of live streamed judo matches. We train models that extract relevant entities and classify combat phases from fixed-perspective judo recordings. We employ semi-supervised methods to address limited labeled data in the domain. We build a model of combat phases via transfer learning from a fine-tuned object detector to classify the presence, activity, and standing state of the match. We evaluate our approach on a dataset of 19 thirty-second judo clips, achieving an F1 score on a $20\%$ test hold-out of 0.66, 0.78, and 0.87 for the three classes, respectively. Our results show initial promise for automating more complex information retrieval tasks using rigorous methods with limited labeled data.