Deep learning-based models have had great success in object detection, but the state-of-the-art models have not yet been widely applied to biological image data. Here we investigate region with convolutional neural network (R-CNN) models based on pretrained networks over natural images to detect nuclei locations in fluorescence microscopic images of autophagy neo-natal ventricular myocytes (NRVM). Cell detection in fluorescence microscopy is a challenging problem due to factors like low contrast, variation in cell shape, colour, and textures. Moreover, factors like image modality and illumination, weak boundaries and overlapping cells, staining and noisy artefacts make most had-crafting cell detection algorithms to fail because they do not generalize well on datasets they have not been developed for or tested on. Our proposed strategy relies on deep neural networks because deep neural networks are said to generalize well across different images modalities and cell types providing that enough annotated dataset is available for training and inference.