Object detection methods using deep learning have achieved significant breakthroughs in recent years. Despite advancements, object detection in rainy weather conditions remains challenging due to limited visibility. In this article, we introduce an enhanced feature-based learning network (EFL-Net) to improve object detection accuracy in the presence of rain. Our proposed model achieves this goal by combining rain and mist removal tasks with object detection task. Specifically, the EFL-Net closely integrates three main components: a deraining subnetwork, a demisting subnetwork, and a detection subnetwork. The deraining and demisting subnetworks are adopted to produce clear features, free of rain and mist, from the degraded input image. In contrast, the detection subnetwork is responsible for locating and classifying objects. Experimental results on both synthesized and real-world datasets prove that the proposed EFL-Net achieves significant improvement in object detection performance, outperforming current advanced competing models.