Many imaging techniques, particularly in environmental transmission electron microscopy (ETEM), generate images with degraded signal-to-noise ratio, contrast and spatio-temporal resolution, which hamper quantification and reliable interpretation of data. Moreover, the extraction of structural information from these images relies on manual acquisition and local structural identification which does not allow statistical analysis of the data and necessarily introduces a human bias carried out at the post-processing stage. The general aim of the ARTEMIA project is to develop a ground-breaking deep learning-based framework for in situ microscopy in liquid and gaseous media allowing the automated, high throughput, real-time acquisition and analysis of ETEM image sequences.Our framework will integrate aberration-corrected in situ ETEM imaging using windowed liquid/gas nanoreactors with denoising and resolution enhancement scheme set up using convolutional neural network (CNN). For model training, datasets consisting of simulated liquid- and gas-phase TEM images will be generated by by atomistic simulations including instrumental noise and imperfections of the microscope optics. In the ARTEMIA project, the CNN models will be applied to the study of two crystalline samples with complementary structural characteristics and electron beam sensitivity, model gold nanoparticles (Au NPs) and microporous zeolite, in reactive gas and/or liquid environments. Our scientific aim will be to gain further mechanistic understanding ofthe growth of model Au NPs in liquid phase and their reactivity in oxidizing and reducing gas environments on one hand and the steaming process of beam-sensitive zeolite on the other hand. The consortium comprises three academic partners (MPQ, LEM, IPCMS) and an EPIC partner (IFPEN) with complementary expertise in liquid and gas ETEM, data science and image processing with special focus on deep learning approaches, atomic modelling and TEM image simulation.