Abstract: A method for detecting defects during laser additive manufacturing, comprising the following steps: B21) a first image is captured, the first image being the image, captured in the infrared spectrum, of an upper surface of a layer of powder exposed to laser scan; B23) the first image is processed using a first convolutional neural network of self-encoding type, in such a way as to produce a defect mask indicating the location of defects at the upper surface of the layer of powder. A method for manufacturing parts during which the presence of defects is detected using the preceding method. A data-processing device, computer program and storage medium for the implementation of this method.