With the advent of artificial intelligence as key technology in modern medicine, surgical data science (SDS) promises to improve the quality and value of the particular domain of interventional healthcare through capturing, organization, analysis, and modeling of data, thus creating benefit for both patients and medical staff. Holistic SDS concepts span the topics of context-aware perception in and beyond the operating room, data interpretation and real-time assistance or decision support. At the same time, minimally invasive surgery using cameras to observe the internal anatomy has become the state-of-the-art approach to many surgical procedures. Contributing to the key aspect of perception, endoscopic vision thus constitutes a central component of SDS and computer-assisted interventions. From this arises the necessity for high-quality common datasets that allow the scientific community to perform comparative benchmarking and validation of endoscopic vision algorithms. With EndoVis, we present you a large collection of publicly accessible datasets comprising various computer vision tasks (classification, segmentation, detection, localization,���) and subdisciplines ranging from laparoscopy to coloscopy and surgical training. These datasets can be used for both de novo development as well as validation of methods. EndoVis organizes highprofile international challenges for the comparative validation of endoscopic vision algorithms that focus on different problems each year at MICCAI, thus representing a major driving force of advancements in the field. This year we propose 5 different sub-challenges under the umbrella of EndoVis: SurgToolLoc - Endoscopic surgical tool localization by leveraging tool presence labels CholecTriplet2022 - Surgical Action Triplet Detection and Localization SAR-RARP50 - Instrumentation segmentation and Action Recognition on robotic Radical Prostatectomy SimCol-to-3D: Simulated Colonoscopy data for 3D (scene) reconstruction SurgT - a challenge for tissue tracking in surgery