A Motor Imagery (MI) based Brain Computer Interface (BCI) system aims to provide neuro-rehabilitation for the motor disabled people and patients with brain injuries (e.g., stroke patients) etc. Motor Imagery is a cognitive process of imagining the movement of a body part without moving it. The aim of this research is to classify the MI tasks by utilizing the occurrence of event related desynchronization and synchronization (ERD\ERS) in the Electroencephalogram (EEG) during these tasks. A set of Complex Morlet Wavelets (CMW) having frequency dependent widths is used to generate high-resolution time-frequency representations (TFR) of the MI EEG signals using a minimal number of channels. A Convolutional neural network is used for classifying the generated TFR. The influence of the value of the number of cycles for the CMW, on the TFR are analyzed in order to give a better classification accuracy for the MI data.