Atrial fibrillation (AF) is a serious cardiovascular disease with the phenomenon of beating irregularly. It is the major cause of variety of heart diseases, such as myocardial infarction. Automatic AF beat detection is still a challenging task which needs further exploration. A new framework, which combines modified frequency slice wavelet transform (MFSWT) and convolutional neural networks (CNNs), was proposed for automatic AF beat identification. MFSWT was used to transform 1-s electrocardiogram (ECG) segments to time-frequency images, then the images were fed into a 12-layer CNN for feature extraction and AF/non-AF beat classification. The results on the MIT-BIH Atrial Fibrillation database showed that a mean accuracy (Acc) of 81.07% from 5-fold cross validation is achieved for the test data. The corresponding sensitivity (Se), specificity (Sp) and the area under ROC curve (AUC) results are 74.96%, 86.41% and 0.88. When excluding an extreme poor signal quality ECG recording in the test data, a mean Acc of 84.85% is achieved, with the corresponding Se, Sp and AUC values of 79.05%, 89.99% and 0.92. This study indicates that it is possible to accurately identify AF or non-AF ECGs from a short-term signal episode.