To solve the problem of low accuracy and low efficiency in manual detection of woven bag defects, an efficient online detection method for woven bag defects is proposed. The method collects images of woven bags online and performs image processing to eliminate interference items and accurately detect defects in woven bags. The image is preprocessed by using the mean filter and gray-scale open and close operations to eliminate black and white stripes and gray-scale unevenness that interfere with defect detection in the image, and reduce noise. Use differential image binarization to perform background segmentation on the image, and extract hole defects, wire drawing defects, and excessive wire gaps, wrinkles, and black objects. At the same time, open and close operation is used to connect the broken defects and eliminate the excessive wire gaps in the silk thread, so as to avoid the omission of small defects. Feature extraction and defect detection are used to eliminate the interference of folds and black objects, and detect holes and drawing defects. Experimental results show that the average correct detection rate of 500 samples reaches 97.20%, the detection efficiency is 720 m/h, and the detection accuracy and efficiency are high.