Object detection is a fundamental task in computer vision that aims to locate and classify objects within an image or video. It plays a vital role in various applications, including autonomous driving, surveillance systems, and robotics. One-stage object detectors such as YOLO models are very fast but they are also less accurate. To improve the performance of YOLO models, the objective of this study is to develop an ensemble model of YOLOv5 models. By combining the outputs of multiple base models, ensemble methods can achieve better accuracy and robustness. The base models used for our proposed ensemble model are YOLOv5n, YOLOv5s, and YOLOv5m. The performance metrics used are mAP, precision, and recall. The experimental results revealed that the proposed ensemble model had mAP is 56.1%, precision of 69.7%, and recall of 42.5%. In relation to the base models used in this study, YOLOv5n had mAP of 48.3% precision of 57.5%, and recall of 37.4%; YOLOv5s had mAP of 53.2%, precision of 64.7%, and recall of 39.1%; and YOLOv5m had mAP of 54.8%, precision of 66.2% and recall of 41.6%. In conclusion, our proposed ensemble model performed well when compared to the base models. The proposed model was able to detect bad roads (pothole roads) that are common in Uganda. Finally, we intend to further our research by integrating the proposed model with continual learning approaches so that the model can improve its detections with time.