Abstract: The invention relates to a method for processing images acquired by an endoscope (12), for the automatic estimation of the position of at least one trocar (58) which is disposed on a wall (56) of the body of a patient, each trocar (58) being configured to permit the passage of an instrument (54) therethrough in order to access a cavity (50) in the body of the patient, characterized in that it comprises a step of acquiring images of the abdominal cavity (50) of the patient using the endoscope (12), each of the images showing at least one instrument (54) passing through each trocar (58), a step of determining, from the images, the position and orientation of the instrument (54) with respect to the position of the endoscope (12), a step of estimating, from the images, the position of each trocar (58) from at least two positions and orientations of the instrument (54).

Abstract: The invention relates to a combined system for selection and annotation of training data of a machine learning model, comprising the model (12) and two databases (14, 16) comprising respectively non-annotated or annotated data, characterized in that the model is configured to select a group of the most useful non-annotated data (20) to be annotated, to machine-annotate the most useful non-annotated data (20), to receive at least one annotation validation and/or correction of the data by an expert (18) in order to provide validated data (24) stored in the second database (16), and to be re-trained by the validated data of the second database (16) for the selection of the most useful non-annotated data and for the interactive annotation.

Abstract: The invention relates to a semi-automatic image segmentation system comprising a neural network (12) configured to receive, at each iteration, input data comprising at least one image (14) to be segmented and to provide output data comprising at least one segmentation mask (16) of each image, characterized in that it further comprises a user interaction loop (18) comprising a user interaction module (20) configured to provide at least one user mask representative of an interaction of a real user and/or of a simulated user on each segmentation mask, said user interaction loop (18) being configured to provide, as input data of the neural network (12), each segmentation mask provided by said neural network coming from a previous iteration of the neural network (12), referred to as past segmentation mask, and the user mask associated with each past segmentation mask.

Abstract: The invention relates to a method for real-time tracking of the placement of a camera with respect to a 3D model comprising a keyframe base comprising a first thread of execution making it possible to estimate the current placement of the camera from a selected keyframe, characterized in that the method further comprises a step of analyzing the quality of the estimation of the placement of the camera from the image stored in a buffer memory, and a step of analyzing the sharpness of the image stored in the buffer memory from data representative of the movement of the camera, and if the quality of the estimation of the placement of the camera from the stored image and the sharpness of the stored image conform to predetermined criteria, a step of adding the image stored in the buffer memory to the keyframe base as a new keyframe.