A key point positioning method, a terminal, and a computer storage medium. The method comprises: upon acquiring a target image, generating, on the target image, according to a pre-set configuration, a target detection region for human facial recognition and positioning; inputting a first target image corresponding to the target detection region into a pre-set first model set, and calculating the position and a first confidence of a first human face key point; when the first confidence is greater than or equal to a pre-set threshold, acquiring a second target image corresponding to the position of the first human face key point in the target detection region; inputting the second target image into the pre-set first model set, calculating a second confidence, and determining the accuracy of the position of the first human face key point according to the second confidence; and when the second confidence is greater than or equal to a pre-set accuracy, determining that the position of the first human face key point is the position of a final human face key point of the target image.