Method for Fault Diagnosis and Location of Television Broadcasting Based on Computer Vision

Artificial intelligence has been widely used in television interviewing, editing and broadcasting; however, bitstream monitoring remains the main broadcast monitoring method. Thus, in this study, a fault diagnosis and localisation method is proposed based on computer vision. The YOLOx algorithm was used to segment television walls into images of separate monitoring points. Subsequently, OCR (Optical Character Recognition) technology was applied to recognise each monitoring node’s name to fulfil the association between names and images. An abnormal television broadcast monitoring image dataset was established using historical abnormal images of a television station. Furthermore, a deep convolutional neural network model was constructed based on the deep learning framework PyTorch to extract features of abnormal images. The model was also used to fulfil fault diagnosis and localisation. The system could also fulfil localisation and intelligently propose fault types and solutions to personnel based on associations of radio and television broadcast streams.