# MonoLoco: Monocular 3D Pedestrian Localization and Uncertainty Estimation

## TL;DR

MonoLoco introduces a lightweight neural network that estimates 3D pedestrian locations and their confidence intervals from monocular images, outperforming existing methods and providing meaningful uncertainty measures.

## Contribution

It presents a novel approach for monocular 3D human localization that predicts confidence intervals, improving accuracy and uncertainty estimation with limited training data.

## Key findings

- Outperforms state-of-the-art on KITTI and nuScenes datasets.
- Surpasses stereo-based methods for distant pedestrians.
- Provides meaningful confidence intervals for localization.

## Abstract

We tackle the fundamentally ill-posed problem of 3D human localization from monocular RGB images. Driven by the limitation of neural networks outputting point estimates, we address the ambiguity in the task by predicting confidence intervals through a loss function based on the Laplace distribution. Our architecture is a light-weight feed-forward neural network that predicts 3D locations and corresponding confidence intervals given 2D human poses. The design is particularly well suited for small training data, cross-dataset generalization, and real-time applications. Our experiments show that we (i) outperform state-of-the-art results on KITTI and nuScenes datasets, (ii) even outperform a stereo-based method for far-away pedestrians, and (iii) estimate meaningful confidence intervals. We further share insights on our model of uncertainty in cases of limited observations and out-of-distribution samples.