Evaluation of potential risks of musculoskeletal disorders in real workstations is challenging as the environment is cluttered, which makes it difficult to correctly and accurately assess the pose of a worker. Most of the traditional motion capture systems cannot deal with these workplace constraints. Being marker-free and calibration-free, Microsoft Kinect is a promising device to assess these poses, but the validity of the delivered kinematic data under work conditions is still unknown. In this thesis we first propose an extensive validation of the Kinect system in an ergonomic assessment context with sub-optimal capture condition. As most of the large inaccuracies come from occlusions, we propose a new example-based method to correct unr...