Cricket Shot Detection from Videos
| Autor: | Ankush Mittal, Jayanta Sharma, Vineet Raj, Archit Semwal, Durgesh Mishra |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
biology
Computer science business.industry Shot (filmmaking) Search engine indexing Feature extraction ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Optical flow Pattern recognition 02 engineering and technology 010501 environmental sciences biology.organism_classification 01 natural sciences Convolutional neural network Support vector machine Cricket - sport Cricket 020204 information systems 0202 electrical engineering electronic engineering information engineering Artificial intelligence business 0105 earth and related environmental sciences |
| Zdroj: | ICCCNT |
| DOI: | 10.1109/icccnt.2018.8494081 |
| Popis: | Classifying various type of bat strokes played in a cricket match has always been an arduous undertaking while indexing the cricket sport. Identifying the type of shot played by the batsman in a cricket match is a substantial aspect as well as one of the unplumbed subjects in this domain. This paper proposes a novel scheme to recognize and classify different types of bat shots played in cricket. The model relies on the state-of-the-art techniques like saliency and optical flow to bring out static and dynamic cues and on Deep Convolutional Neural Networks (DCNN) for extracting representations. Moreover, a completely new dataset of 429 videos, has been introduced to evaluate the performance of the proposed framework. The model achieves an accuracy of 83.098% for three classes of right-handed shots and 65.186% for three classes of left-handed shots. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|