# Remote Estimation of Free-Flow Speeds

## TL;DR

This paper introduces a deep learning method to estimate free-flow speeds of roads using overhead imagery and minimal road attributes, reducing reliance on costly detailed labels.

## Contribution

It presents a CNN-based approach that effectively estimates free-flow speeds from imagery alone or combined with simple road features, improving estimation accuracy.

## Key findings

- Imagery alone nearly matches the accuracy of detailed road features.
- Combining imagery with road features yields the highest estimation accuracy.
- The method performs well on a large, diverse dataset.

## Abstract

We propose an automated method to estimate a road segment's free-flow speed from overhead imagery and road metadata. The free-flow speed of a road segment is the average observed vehicle speed in ideal conditions, without congestion or adverse weather. Standard practice for estimating free-flow speeds depends on several road attributes, including grade, curve, and width of the right of way. Unfortunately, many of these fine-grained labels are not always readily available and are costly to manually annotate. To compensate, our model uses a small, easy to obtain subset of road features along with aerial imagery to directly estimate free-flow speed with a deep convolutional neural network (CNN). We evaluate our approach on a large dataset, and demonstrate that using imagery alone performs nearly as well as the road features and that the combination of imagery with road features leads to the highest accuracy.