In this paper, Near infrared spectrum (NIR) technology is used to establish a clothing material and content detection model. Selecting 8 kinds of clothing samples and cotton-polyester blended fabrics as test objects, the detection model of clothing material and content was established through data preprocessing, feature extraction, classification recognition and regression analysis. The experimental results show that the qualitative discriminant model established by the least square vector machine has higher accuracy and shorter operation time. The quantitative analysis model based on the NIR data of the cotton-polyester mixed fabric after the second derivative smoothing and pretreatment combined with the partial least square algorithm has good accuracy and stability, and the predicted root-mean-square error reaches 0.0019, which meets the accuracy requirements of the NIR rapid detection. The near infrared detection method established in this paper has the advantages of environmental protection, short time and low cost, and has certain application value in the fields of clothing manufacturing, quality assurance and quality inspection. This study provides a new method and idea for the detection of fabric material and content, which has certain theoretical and practical application value.