The "Lettres en Lumières" project aims to create a collaborative platform dedicated to the transcription and segmentation of historical documents. This platform has enabled the annotation of the DRoSB base, which combines manual segmentation data for training object recognition models, such as Mask-RCNN, and transcription data initially obtained through semi-automatic processes. Correction of the transcriptions, performed by volunteers and validated by professional palaeographers, ensures the accuracy and reliability of the data. The platform offers several interfaces to manage these tasks, including segmentation correction, transcription, transcription validation, and management. In addition, "Lettres en Lumières" uses the IIIF format to publish the results and allow for a wide dissemination of the transcribed documents. This format facilitates the sharing and visualization of high-resolution images and textual and graphical annotations. The project has been designed with a multi-tenant architecture that guarantees the autonomy and confidentiality of each partner institution, while allowing the platform to be continuously updated to meet users' needs.