Image classification based on Alexnet-Tensorflow and SVM. Please understand that original ECG images are removed because of privacy. For test images you could use corns images examples in https://github.com/QiyuanMa/Image-classification-SVM-ML The models and codes are the same.
This project provides a image classification model based on Alexnet-tensorflow and compared it with SVM. The accuraty is 0.87, but you should adjust the parameters for your images.
Run run.py, you should run train(), generate_pre_result(), evaluation_eval_dataset() individually. And evaluation_eval_dataset is only for result visualization.
(1)The generated txt, which is for providing images locations for further steps.
(2)Ttraining models in checkpoint, please ba aware that the model selection should be effected by best_val_acc. You should use the model which testing accuracy higher than best_val_acc.
(3)Training results(study rate is 0.0001)
(4)Model labels(run test_pred_labels)
Run evaluation_eval_dataset()
(1)The above 5 is correctly classified images,the below 5 is wrong classified images.
(2)The comparation of correct and wrong images.
(3)The SVM image classification final accuracy result
(4)The Alexnet image classification final accurary result
