In this study, AI Scout and AI Data Recorder systems automatically cutting the pitch-by-pitch clips and record the plate appearance results are developed. With appropriate tuning and training, our systems are expected to be applied onto the professional baseball leagues with online game broad casting for complete player scouting. Based on transfer learning using InceptionResNetv2, our AI Scout can automatically the pitch-by-pitch clips from the online baseball game broadcasting and reduce all human efforts of film cutting and 85% of human effort in clip review. 

Both professional and national baseball teams can be benefited from our system in complete information gathering and dramatic human efforts reducing. The experts in scouting and information analysis can be relieved from the cumbersome work of film cutting and focus on matters more important.

Our methodology integrates several computer vision technology based on Convolutional Neural Network to relieve the huge manpower needs on professional baseball scouting. The prototype which can be applied on domestic professional baseball broadcasting has been completed, and can be directly applied for baseball player scouting and information collection in domestic professional baseball league. Once more data and labels are collected, we are confident that this technique can be applied worldwide.

To the best of our knowledge, there＇re no similar and published technique around the world. Thus, our methodology has great potential in commercial application.

The system developed in this study can dramatically (90%) reduce the efforts necessary to baseball scouts in film editing and data recording. The system can gather the information and intelligence of professional baseball player from the leagues with public broadcasting worldwide. The professional or national baseball team can therefore be benefited.