Dynamic fragmentation is a fast and catastrophic failure of a material, induced by extreme loads or as a consequence of a material instability. The nucleation, propagation, branching and coalescence of cracks results in the formation of a number of fragments which depend on the loading rate. Being able to predict the fragmentation pattern and the history of fragments formation for a given loading case is thus of crucial importance in various industrial fields, such as the design of armoured structures, mastering the number of generated spatial debris in aerospace engineering, etc. Such objective requires to address two main challenges. The first one is the construction of well-posed mathematical modellings and associated efficient 3D numerical solvers, whose predictions can be validated against quantitative experimental data. The second one is to get a better understanding of the complex crack response when interacting with mechanical waves, and to provide exhaustive experimental data which will be highly discriminant for testing models. The IPERFrag project aims at performing an image-based assessment of a new 3D hyperbolic gradient damage modelling to describe the fragmentation in brittle solids. This project thus aims to provide contributions to the two above main challenges. First, the proposed modelling finds its originality in that it embeds a damage micro-inertia leading to the description of the damage propagation through a wave. It thus allows taking advantage of existing finite volume schemes, of lower complexity than existing non-local damage solvers. Its implementation within the open source computer code ECOGEN will permit to envisage truly 3D industrial applications in the medium term, especially through high performance computating (HPC) capabilities. Second, high-spatial resolution (HR) and ultra-high speed (UHS) imaging combined with digital image correlation (DIC) will be carried out to accurately capture the interaction between the fragmentation pattern and mechanical waves, thanks to available multi-sensor rotating mirror camera. Experimental dynamic tests conducted on discriminant testing configurations, involving an increasing complexity of fragmentation patterns (from 1D to 3D), will allow getting quantitative in-situ data through full-field measurements with unprecedented resolution. Among the latters, can be cited crack front self-emitted acoustic waves, crack/wave interactions, branching conditions, fragment kinematics and morphology depending on the loading. The last step of the IPERFrag project consists in calibrating the proposed hyperbolic gradient damage modelling with respect to a subset of experimental data obtained on quasi-brittle Poly(methyl methacrylate) (PMMA) material, and to test its predictive capabilities on the complementary one, including a challenging 3D plate impact. In a nutshell, the IPERFrag project aims at providing the scientific community with (i) a calibrated and efficient free (open-source) 3D computer code able to address true 3D industrial fragmentation scenarii of brittle solids, (ii) unique quantitative data from increasing complexity experimental dynamical tests up to 3D fragmentation, and (iii) a comprehensive in-situ visualisation of the underlying mechanisms of formation of fragments and their associated chronology.