TEXT RECOGNITION METHOD AND TERMINAL DEVICE

A text recognition method, comprising: performing feature extraction on an image to be recognised to acquire a first feature map; processing the first feature map to at least acquire N first candidate vector detection frames, each first candidate vector detection frame being configured to frame a text vector area; filtering the N first candidate vector detection frames to acquire K first target vector detection frames; performing feature extraction on the first feature map to acquire a second feature map; processing the second feature map to acquire L first candidate text detection frames, each first candidate text detection frame being configured to frame an area containing text; on the basis of the K first target vector detection frames, filtering the L first candidate text detection frames to acquire J first target text detection frames; and recognising the text in the J first target text detection frames to acquire J items of target text information. La présente invention concerne un procédé de reconnaiss