PCB-TD：PCB image Dataset

     This is a PCB image dataset containing text detection bounding boxes. Our dataset comes from dataset given by Li Wei[1] and from the website such as TaoBao, JingDong, etc.
        As shown in Table 1, the training set consists of 298 images with different PCB scales and 55647 annotated targets are present. The number of targets in each image varies from tens to hundreds and the size of these images is also different. The testing set includes 50 PCB images with dozens of targets in each image. Notably, the testing set contains the PCB image that are not in training set, which increases the difficulty in the task and challenges the scalability of the deep model.         This dataset contains different images of two scales:
        (1) Small scale images and annotations         (2) Large scale images and annotations

Reference:
        [1] Li W , Neullens S , Breier M , et al. Text recognition for information retrieval in images of printed circuit boards[C]// Conference of the IEEE Industrial Electronics Society. IEEE, 2014.

Download PCBTD Dataset