Glass tube liquid level meter identification method and related device

The invention discloses a glass tube liquid level meter identification method and a related device. The method comprises the following steps: acquiring a first image containing a glass liquid level meter, and inputting a trained posture identification model to determine a plurality of groups of key points corresponding to the first image; selecting a second image only containing the glass liquid level meter from the first image based on the multiple groups of key points, and performing image segmentation to obtain a liquid level mask image of the glass liquid level meter; and determining the liquid level reading of the glass liquid level meter based on the second image and the liquid level mask map. According to the method, the accuracy and the speed of key point identification are improved by processing the convolutional neural network of the first image and cooperating with an attention mechanism, and the accurate segmentation of the liquid column of the liquid level meter is improved by adopting an image c