The topic of this research is related to the introduction of a new artificial intelligence-based approach to control repetitive production in denim garment manufacturing. Thus, the similarity and repeatability of productions in repeated orders can be easily determined. The focus is on generating an output based on the similarity of input images. This is done by determining whether the images resulting from the repetitive production of the same products belong to the same product. There are some elements that make this work more difficult than classical image comparison. These include: (i) the images may have been taken under different environmental conditions, (ii) the images may have been taken at different zoom levels, (iii) not all of the product may be visible in both images (iv) the effects on the denim garment. To solve these problems, a deep learning based integrated control system is introduced in addition to the image processing approach. A four-step methodology was used in the study. First, the images were pre-processed using image processing methods. After pre-processing, transfer learning-based segmentation was performed to separate the product parts in the images. Convolutional neural network approaches were used to extract features for each product part. The final similarity ratios were calculated and the similarity ratios of the images were calculated using different similarity measurement methods. As a result of the evaluation on 284 sample images taken from production, it was observed that the proposed approach was able to make predictions with an accuracy of approximately 80% according to pairwise comparisons.