Sports Video Classification Framework Using Enhanced Threshold Based Keyframe Selection Algorithm and Customized CNN on UCF101 and Sports1-M Dataset

Table 1

Summary of the related work.

Ref.	Year	Title	Model	Optimization technique/algorithm	Dataset	Class and accuracy	Outcome
[9]	2021	Olympic Games event recognition via transfer learning with photobombing guided data augmentation	AlexNet, VGG-16, ResNet-50	Transfer learning	OGED - Olympic games event image dataset	Multiclass and 90%	Olympic Game event recognition
[10]	2021	Categorization of actions in soccer videos using a combination of transfer learning and gated recurrent units	CNN, RNN, and soccer actions categorization	—	SoccerAct10	94%	10 soccer actions corner, foul, free-kick, goal-kick, long-pass, penalty, and so on.
[8]	2021	Sports recognition using convolutional neural networks with optimization techniques from images and live streams	Extended Resnet50 and VGG16	RMSProp, ADAM & SGD	5sports	Resnet50-83% and VGG16-95%	Sports event recognition.
[5]	2021	Traditional Bangladeshi sports video classification using deep learning method	CNN and LSTM	—	Traditional Bangladeshi sports video (TBSV), UCF sports, UCF101	5 classes and 99%	Bangladeshi sports Vido classification.
[3]	2021	A sports training video classification model based on deep learning	AlexNet		Various dataset	9 classes and 99%	Sports training video classification.
[4]	2021	Deep learning for video classification: A review	2D-CNNs, 3D-CNNs, handcrafted approaches.	—	—	—	Video classification in general
[2]	2020	Activity recognition framework in sports videos	Deep learning	K-means clustering	YouTube, cric-info	Multiclass	Frames extracted
[6]	2020	SSET: a Dataset for shot segmentation, event detection, player tracking in soccer videos	DevNet, VGG LSTM replay, LRCN, GoogLeNet	—	Soccer Dataset for Shot, Event, and Tracking (SSET)	Multiclass	Shot segmentation, event detection, player tracking
[1]	2020	Video event classification based on two-stage neural network	CNN and RNN	Transfer learning	UCF101, HMDB51 and CCV	Multiclass	Video event classification
[11]	2020	A K-means clustering approach for extraction of keyframes in fast-moving videos	—	Shot boundary detection, keyframe extraction		Three classes 23.52%, 14.11%, and 6.62%	Keyframe extraction
[12]	2019	Shot classification of field sports videos using an AlexNet convolutional neural network	AlexNet with the proposed framework	—	ESPN, star sport, sky sports, ten sports, etc.	Multiclass 94.07%	To classify the shots into long, medium, close-up, and out-of-the-field shots.
[13]	2019	Video genre identification using clustering-based shot detection algorithm	SVM, CNN	K-mean, K-medoid		Two classes and 90%	An audio talk show or another video
[14]	2019	Keyframe extraction based on HSV histogram and adaptive clustering	—	K-means, density peak clustering algorithm (DPCA), partition based clustering, I-frame	—	—	Keyframes
[7]	2018	Keyframe extraction for video summarization using local description and repeatability graph clustering	—	Graph clustering	Open video project (OVP) and YUV video sequences	—	Video summarization
[15]	2015	Real-time event classification in field sport videos	—	Decision tree, feed-forward neural network	Ground truth dataset together with an annotation technique	—	Event identification