Aiming at the problem of low detection accuracy caused by small volume, dense distribution and complex background of aircraft targets in the color remote sensing images, an improved aircraft target detection algorithm in any direction of the color remote sensing images based on anchor-free is proposed. Using BBAVectors as the benchmark model and ResNet50 as the backbone network for feature extraction. after the feature pyramid network (FPN), a top-down path augmentation network (PANet) module is added to shorten the information path and enhance the feature pyramid with low-level accurate location information. Secondly, the attention mechanism convolutional block attention module(CBAM) is introduced to improve the accuracy of aircraft target detection in complex environment by suppressing the noise and highlighting target characteristics. Ablation experiments and comparative experiments are conducted on DOTA data sets, and DOTA_ Devkit is used to cut the data set by 0.5 and 1 times respectively to improve the detection accuracy of the model. The detection accuracy of the improved model on the color remote sensing image test data set reaches 90.35%. Compared with the original model, the detection accuracy is improved by 0.82%. The experimental results show that this method has better detection effect in the aircraft detection task in color remote sensing images.