Insulator defect detection is a crucial aspect of power line inspection. To achieve effective detection of insulator defects, a deep learning-based object detection method is applied to unmanned aerial vehicles, offering advantages such as accuracy, efficiency, and low cost. While the You Only Look Once Version8 network demonstrates superior performance in insulator defect detection using unmanned aerial vehicles imagery compared to other methods, it still struggles to achieve ideal results in scenarios with variable target scales and complex backgrounds. To address this issue, an improved network is proposed in this paper, tailored to the characteristics of insulator defect detection using unmanned aerial vehicles imagery. Firstly, an attention mechanism is introduced into the Backbone network of our network to attenuate the influence of background in the target region. Secondly, the bidirectional feature pyramid network structure is incorporated, and cross-layer connections and weighted fusion are implemented during feature extraction and fusion to enable the network to better focus on insulator defect features and suppress the impact of noise. Lastly, the detection head at the lowest layer is replaced with a small target detection head in our network to enhance the network's attention to small target defect areas. Experimental results on a self-made dataset demonstrate that recall and precision of our network are 89.9% and 96.5%.