|
| Author | Proposed or tested model | Categorize | Dataset | Evaluation |
|
| Jing et al. [89] | Mobile-Unet | One-stage | Benchmark databases, the fabric images database (FID), and yarn dyed fabric images (YFI), in which all images are manually annotated segmentation | Pixel accuracy (PA) and IoU |
|
| Hong-wei hang [90] | YOLOV2 | One-stage | Collected dataset (276) | IOU, recall, and precision |
|
| Young-Joo Han [91] | Stacked convolutional autoencoders | One-stage | Synthetic and collected dataset | Recall, precision, and F-score |
|
| Xinying He [92] | Adaptive method based on DenseNet-SSD | One-stage | Collected dataset (2072) | Calculate localization loss (loc) and confidence loss (conf) |
|
| Mohammed et al. [93] | A multilayer perceptron with a Levenberg–Marquardt (LM) algorithm | One-stage | Collected dataset (217) | Specificity, accuracy, and sensitivity |
|
| Shuang mei [94] | Multiscale convolutional denoising autoencoder network model | One-stage | Four datasets: fabrics, KTH-TIPS, Kylberg texture, and ms-texture | Recall, precision, and F1-measure |
|
| Huosheng Xie [95] | Improved RefineDet | One-stage | TILDA dataset, Hong Kong patterned textures database, and DAGM2007 dataset | Precision (P), recall I, F1-score, mean average precision (mAP), model parameter (param.) |
|
| Yanqing Huang [96] | Segmentation network and decision network | Two-stage | Dark redfFabric (DRF), light blue fabric (LBF) and patterned texture fabric (PTF) | Frames per second (FPS) Avg-IoU and Avg-P |
|
