Computer vision enables computers to interpret and analyze visual information. In investing, it turns satellite and webcam imagery into movement data, such as counts of vehicle brands in specific areas. section 11.1.1
Its value depends on both signal and execution. Training models to identify objects correctly requires substantially more effort than analyzing mainstream data, but underrepresented visual data may provide investors with an advantage. section 11.1.1
Common applications include:
- Estimating the growth of autonomous-vehicle deployments by counting cars in selected locations. section 11.1.1
- Assessing whether computer vision can replace LiDAR in autonomous driving systems, with major implications for sensor costs and suppliers. section 4.3.1
