Motor imagery on electroencephalogram (EEG) signals is widely used in brain-computer interface (BCI) systems with many exciting applications. Recently, many deep learning classifiers have been adopted, especially Convolutional Neural Networks (CNNs) in BCI application. However, CNNs suffer from the loss of salient features during training, causing the spatial invariant problem that affects the performance. This study develops a framework using CSP and Short-Time Fourier Transform (STFT) with Attention Convolutional Neural Network (CSP-STFT CNN) for EEG BCI classification. The features from CSP are translated into the spatial domain using STFT as input to attention-based CNN as the classifier. This framework uses attention-based CNNs to classify the collected spatial images across different test subjects. Finally, the performance of the CSP-STFT CNN is validated on BCI benchmark datasets, Competition III dataset Iva. The proposed CSP-STFT offers a promising result; the classifier achieved better performance in terms of classification accuracy, averaging 80% across all five subjects for Competition III dataset IVa. The precision and recall are excellent too, ranging around 0.8-0.9. In general, the proposed CSP-STFT CNN can offer richer joint spatiotemporal features as inputs to classifiers, whereas using an Attention-CNN classifier improves upon the earlier problems suffered by CNNs.