Electrocardiographic imaging (ECGI) aims to noninvasively estimate heart surface potentials from body surface potentials, and torso and heart geometries. Practical challenges and costs associated with its implementation have limited its application as a wider screening tool. In this study, we developed a deep learning framework to estimate heart surface potentials solely from body surface potentials. The framework introduces three main components: the transformation of 3D torso and heart geometries into corresponding standard 2D representations, and the development of a customized deep learning model. The standardized 2D torso and heart representations are generated to maintain a consistent layout across different subjects, while preserving the subject-specific anatomical characteristics. The deep learning model is based on a Pix2Pix network, adapted to work with 2.5D data in our task, i.e., 2D body surface potential maps (BSPMs) and 2D heart surface potential maps (HSPMs) with time sequential information. BSPMs and HSPMs from 11 healthy subjects (8 females, 3 males) and 29 idiopathic ventricular fibrillation (IVF) patients (11 females, 18 males) were used in this study. Performance was assessed by comparing the outcome of the deep learning model with the reconstructions provided by standard ECGI. Results on the testing set show that the mean of the mean absolute error (MAE) for the HSPMs was 0.012±0.011, and the mean of the corresponding structural similarity index measure (SSIM) was 0.984±0.026. The mean of the MAE for the concatenated electrograms (EGMs) was 0.004±0.004, and the mean of the corresponding Pearson correlation coefficient (PCC) was 0.643±0.352. The mean of the time differences between estimated and reference activation times was 1.844±7.775ms, and -4.164±25.183ms for recovery times. Results show comparable performance to standard ECGI without the need for CT/MRI, thus suggesting the possible use of this framework in cost effective clinical scenarios like patient screening or post-operative follow-up.