Detection Of Snowball and Jersey Colour And Jersey Number Using CNN
D C Shubhangi, Baswaraj Gadgay, Mumtaz Begum, MA Waheed · 2023
A snowball is spherical item composed of snow that is often formed by scooping snow into the hands and compacting it into a ball. Games like snowball battles frequently involve the use of snowballs. A larger snowball can also be created by rolling a smaller snowball across a snowy surface. As it rolls, the smaller snowball gathers more snow, which causes it to grow. Tracking separate performers on ice as well as team strategic decision-making depend on the ability of coaches and other analysts to accurately identify snowball performers and squads in a game. In contrast, snowball is a fluid sport because of the complicated situations and frequent team substitutions, which cause the players to adopt different postures while playing. In this proposed system, individual ice snowball players are distinguished using a 2-phase cascaded convolution neural network (CNN) prototypical, jersey color of discovered individuals is recovered to additional ascertain team connections. Because we used a dataset we created from real images, our algorithm is able to remove most of disturbing details, which results in an overall detection accuracy as well as recall amount of over 95% for separate performers, an average detection accuracy of over 96% for teams. With our own dataset, we are able to reach maximum degree of accuracy in our player detection algorithm.