Table 24. Comparison of accuracy between the proposed system and existing systems.
| Authors | Models | Dataset | Batch Size | Epochs | Optimizer | Learning Rate | Accuracy |
|---|---|---|---|---|---|---|---|
| Ameri A. | CNN | HAM10000 dermoscopy image database (3400 images) | 30 | 40 | SGDM | 0.0001 | 84% |
| Lequan Yu et.al. | FCRN | ISBI 2016 Skin Lesion Analysis Towards Melanoma Detection Challenge | 4 | 3000 | SGD | 0.001 | 85.5% |
| Andre Esteva et. al. | CNN | The Stanford University Medical Center and undisclosed Online Databases (129,450 clinical images) | - - - - | - - - - | - - - - | - - - - | 96% |
| Shunichi Jinnai et. al. | FRCNN | 5846 clinical images from 3551 patients. | 4 | 100 | SGD | - - - - | 91.5% |
| Joakim Boman, et. al. | CNN | ISIC Dermoscopic Image Dataset (23,647), DermQuest (16,826), Dermatology Atlas (4,336), a total number of 1,948 images had been taken from DermaAmin, Dermoscopy Atlas, Global Skin Atlas, Hellenic Dermatological Atlas, Medscape, Regional Derm, Skinsight and the pH2 database. | 100 | 30 | - - - - | 0.001 | 91% |
| Rehan Ashraf et. al. | CNN | DermIS: 3176 images DermQuest: 1096 images |
- - - - | 10 | - - - - | Small rate | 97.9% 97.4% |
| Manu Goyal et. al. | Faster-RCNN | International Symposium on Biomedical Imaging ISBI—2017 testing dataset | - - - - | - - - - | ——- - | - - - - | 94.5% |
| Abder-Rahman Ali et. al. | F-MLP | ISIC 2018: Skin Lesion Analysis Toward Melanoma Detection” grand challenge datasets | - - - - | 20 | FGD | 0.001 | 95.2% |
| Yasuhiro Fujisawa et. al. | CNN | ILSVR2012 dataset: Containing 1.2 million images within 1,000 classes | 30 | -—-—- | - - - - | - - - - | 92% |
| Proposed model SCNN_12 |
CNN | Kaggle skin cancer image ISIC archive dataset consisting of 3297 skin cancer images (1800 Benign and 1497 Malignant). After augmentation: 16485 images | 64 | 200 | Adam | 0.001 | 98.74% |