---
updatedAt: 2026-08-27T14:41:47.000Z
agentTools:
  projectIndex: https://documentation.picsellia.com/llms.txt
---

# Datalake

The `Datalake` is where every image and video you bring into Picsellia physically lives, independently of any `Dataset` or `Model` built on top of it.

For Computer Vision teams, keeping raw `Data` in a `Datalake` before it is annotated is what makes it possible to store a file only once and reuse it across several `Dataset`, structure and search through large, unlabeled volumes of images and videos, and plug Picsellia directly into an existing storage pipeline instead of duplicating data.

This section covers how to bring `Data` into a `Datalake`, organize and explore it, and connect it to your own cloud storage, starting with [Philosophy and infrastructure](https://documentation.picsellia.com/docs/datalake-3).