The algorithm based on infrared target detection plays an important role in military and civil fields. However, under the condition of low signal-to-noise ratio, it is difficult to detect infrared spotted targets such as aircraft and ships, with high false detection rate and false alarm rate.?To solve this problem, an algorithm based on machine learning to detect infrared speckled targets with pixel areas ranging from 3×3 to 100×100 is proposed.?In the algorithm part, morphological method is used to pre-extract the target, and HOG feature extraction and SVM machine learning are used to classify the real target.?The proposed algorithm achieves 94.01%, 92.86% and 92.19% detection accuracy in large, medium and small scale target detection respectively.?In addition, the algorithm on SoC platform is implemented, and the algorithm has high real-time performance on the basis of low resource utilization.