This paper presents a novel lightweight model called Leaner YOLOv7-Tiny for detecting transmission line faults from aerial images. The primary goal is to address the issues of high parameter complexity and computational load in existing fault detection algorithms, which hinder their deployment on devices like drones. The proposed model retains the ELAN structure from YOLOv7-Tiny but replaces its backbone with depthwise separable convolutions to reduce model parameters. It also integrates the SP attention mechanism to capture multi-scale features and enhance small target recognition. Additionally, an improved FCIoU Loss function is introduced to balance the contribution of high-quality and low-quality samples, accelerating model convergence and boosting detection accuracy. Experimental results show a 20% reduction in model size compared to the original YOLOv7-Tiny, with detection accuracy surpassing current mainstream lightweight object detection algorithms. The approach holds practical significance for transmission line fault detection, particularly in drone-based inspections.This paper presents a novel lightweight model called Leaner YOLOv7-Tiny for detecting transmission line faults from aerial images. The primary goal is to address the issues of high parameter complexity and computational load in existing fault detection algorithms, which hinder their deployment on devices like drones. The proposed model retains the ELAN structure from YOLOv7-Tiny but replaces its backbone with depthwise separable convolutions to reduce model parameters. It also integrates the SP attention mechanism to capture multi-scale features and enhance small target recognition. Additionally, an improved FCIoU Loss function is introduced to balance the contribution of high-quality and low-quality samples, accelerating model convergence and boosting detection accuracy. Experimental results show a 20% reduction in model size compared to the original YOLOv7-Tiny, with detection accuracy surpassing current mainstream lightweight object detection algorithms. The approach holds practical significance for transmission line fault detection, particularly in drone-based inspections.