Abstract: Conventional drill pipe loading and unloading systems mainly rely on mechanical positioning and lack target following capability. To address this issue, this study proposes a Faster R-CNN-based drill pipe localization method using Differential Evolution (DE). The method obtains the three-dimensional spatial coordinates of drill pipes in the drill pipe box through machine vision, thereby supporting the automatic loading and unloading of drill pipes in intelligent drill rigs. Using top-view camera images of drill pipes, the study builds a DE-Faster R-CNN model. A partial convolutional structure of MobileNetV2 is adopted as the feature extraction backbone, and the differential evolution algorithm is employed to optimize the batch size and the momentum coefficient of the SGD optimizer. On this basis, a depth camera is used to measure the target depth information in real time, enabling dynamic acquisition of the three-dimensional spatial coordinates of the drill pipes. Experimental results show that the proposed method achieves a localization accuracy of 10 mm. Compared with existing methods, it offers higher accuracy and faster execution speed, meeting the localization requirements of drill pipe loading and unloading systems. The DE-Faster R-CNN- based drill pipe localization method features high accuracy, a simple structure, and adaptive parameter tuning, providing strong support for the realization of automatic drill pipe loading and unloading technology in intelligent drill rigs.