Natural initiation of pineapple flowers is not synchronized, which yields difficulties in yield prediction and the decision of harvest. Computer vision based pineapple detection system is an automated solution to address this issue. However, it is faced with significant challenges, e.g. pineapple flowers and fruits vary in size at different growing stages, the images are influenced by camera viewpoint, illumination conditions, occlusion and so on. This paper presents an approach for pineapple fruit and flower recognition using a state-of-the-art deep object detection model. We collected images from pineapple orchard using three different cameras and selected suitable ones to create a dataset. The experimental results show promising detection performance, with an mAP of 0.64 and