Table 1. A summary of the literature review showing the past techniques and their limitations.

Authors	Task	Models	Limitations
Ameri et. al., [14]	Classification, segmentation	CNN	i. Further experimentation of proposed model’s parameters is absent. ii. Absence of image pre-processing and data augmentation technique
Manu Goyal et. al., [15]	segmentation	Mask R-CNN DeeplabV3+	i. Lack of ablation study on proposed model ii. Lack of image processing and data augmentation techniques that might have given better accuracy
Kharazmi et. al., [16]	Classification	Feature extraction Random forest	i. No eliminatation of artefacts (bubble) that are present in the images ii. Absence of ablation study in proposed model
Albahar et. al., [17]	Classification	CNN model with novel regularizer	i. Lack of ablation study in proposed model
Sikkander et. al.. [18]	Segmentation Classification	ANFC	i. Experimentations with other deep learning models is absent.
Sagar et.al., [20]	Classification	CNN	i. Absence of image preprocessing techniques ii. Use of a specific optimizer and learning rate
Ashraf et. al., [21]	Segmentation	YOLOv4	i. Lack of artefacts removal techniques ii. Use of a specific optimizer and learning rate
Wei et. al., [23]	Classification	CNN	i. Size of input image is too large requiring higher resources ii. Experimentations with various loss functions is absent