Dense Fog is a large multimodal dataset for object detection. The dataset includes 12,000 samples from real-world driving scenarios and 1,500 samples from controlled weather conditions in a fog chamber. The dataset includes different adverse weather conditions such as fog, snow, and rain, collected over 10,000 kilometers of driving in Northern Europe. The dataset has a total of 100,000 objects annotated with accurate 2D and 3D bounding boxes.