<p>With the advancement of technology, an automatic oil palm trees recognition can be made possible using a deep learning object detection model. Nevertheless, research on the detection and recognition of both young and mature oil palm trees using single-stage object detectors remains inadequate. Furthermore, these deep learning models often function as black boxes, lacking interpretability and transparency in their decision-making processes. To address this challenge, we proposed a novel and enhanced model, namely YOLOv7e6e-RepVGGN for recognising both young and mature oil palm trees. This approach builds upon the success of the existing YOLOv7 model by incorporating enhancements to achieve accurate recognition. The proposed modification was tested in one small dataset and one public complex dataset, and it surpassed YOLOv7 and the other state-of-the-art models such as YOLOv5, YOLOv6, Single Shot MultiBox Detector (SSD), and Faster-RCNN with the highest average precision and average recall of 97.50% and 99.60% respectively. Moreover, our proposed methods achieved an average recall of 92% at an Intersection over Union (IoU) threshold of 0.75, outperforming other methods, which recorded recall values ranging from 85% to approximately 90%. In addition, we further deployed Eigen-CAM, Ablation-CAM, and Score-CAM in the category of Gradient-free CAM algorithms to provide the visual explanation and interpret the prediction. Our proposed method demonstrated its ability to assess the maturity of oil palms, offering potential for its integration into the oil palm plantations management.</p>