Tunnels in hard rock are typically supported with a thin layer of fibre-reinforced shotcrete in combination with rock bolts. Cracks in the shotcrete could lead to corrosion of the fibres, which reduces the residual strength and could lead to downfall of shotcrete. Therefore, routine inspections are carried out to maintain a safe tunnel. Today, visual inspections are mainly performed, which is timeconsuming and prone to human errors. TACK (Tunnel Automatic CracK Detection) is a research project that aims to develop an autonomous tunnel inspection method based on a hybrid approach of photogrammetry and deep-learning. Data from the tunnel is collected with a mobilemapping system equipped with LiDAR sensors and high-resolution cameras. LiDAR data is used to create a 3D model of the tunnel. Then, a deep-learning approach is used to automatically detect cracks in the acquired images. Once the cracks are detected, a novel photogrammetric algorithm is used to calculate the geometric features of the cracks, i.e. length and width. Finally, the risk associated with the cracks is assessed, and critical sections in need of repair or visual inspections can be pointed out. This paper presents a case-study based on data collected from one tunnel.