DL4MicEverywhere: deep learning for microscopy made flexible, shareable and reproducible

Deep learning enables the transformative analysis of large multidimensional microscopy datasets, but barriers remain in implementing these advanced techniques. Many researchers lack access to annotated data, high performance computing (HPC) resources and expertise to develop, train and deploy deep learning models. In recent years, several approaches have been developed to democratize deep learning usage in microscopy. Tools such as the BioImage Model Zoo facilitate sharing and reuse of pretrained models, distributing them as one-click image processing solutions. Yet often, deep learning models need to be trained or fine-tuned on the end-user dataset to perform well. We previously released ZeroCostDL4Mic, an online platform relying on Google Colab that helped democratize deep learning by providing a zero-code interface to train and evaluate models capable of performing various bioimage processing tasks, such as segmentation, object detection, denoising, super-resolution microscopy and image-to-image translation. Here, we introduce DL4MicEverywhere, an advancement of the ZeroCostDL4Mic framework.