Combining visual motion and luminance features to enhance the detection of small moving objects in a bioinspired model

TL;DR

This paper proposes a bioinspired model that combines visual motion and luminance features to improve the detection of small moving objects, inspired by the visual systems of flying insects.

Contribution

The study introduces a novel fly-inspired model that integrates visual motion and luminance features for enhanced small moving object detection.

Findings

The model exhibits hyperacute object detection capabilities not typically associated with motion detection alone.

Combining motion and luminance features improves detection efficiency and robustness in realistic scenarios with moving backgrounds.

The model outperforms existing insect-inspired methods on real-world datasets, including infrared drone detection.

Abstract

Flying insects demonstrate exceptional proficiency in detecting and pursuing conspecifics and prey within a cluttered environment, inspiring the development of computational models for small object detection. While existing bioinspired models are dedicated to resolving small moving instead of stationary object detection, few studies have systematically explored the role of visual motion in detection. Here, we developed a fly-inspired model on the basis of the hypothesis that combining visual motion features and luminance features is critical for small moving object detection. We thoroughly investigated the effect of feature combination under diverse stimulus conditions. Simulations indicated that the model exhibited hyperacute object detection, a capability not generally believed to emerge on the basis of motion detection. When tested with a moving background in realistic scenarios, the model demonstrated enhanced efficiency and robustness relative to models relying solely on luminance features. This enhancement was independent of whether visual motion was extracted by two- or three-arm motion detectors. The results suggested that small object detectors within the visual systems of flying insects could be optimally tuned to utilize the limited features inherent to tiny objects.

Flying insects are adept at detecting small moving objects on a daily basis under natural conditions, even if these objects are smaller than the spatial resolution of their compound eyes. In contrast, small object detection is still a challenge for artificial vision systems, such as those used for robots, even with the help of deep convolutional neural networks. The main challenge is that tiny objects have few visual features. Thus, it is valuable to draw inspiration from the neural algorithms of miniature insect brains, which resolve this issue by relying on the limited features of a tiny object. While a variety of neurons selectively responding to the movement of small objects have been identified in fruit flies, hoverflies, and dragonflies, little is known about whether the neurons specifically responsible for detecting visual motion are involved in small moving object detection. Systematic investigations of how visual motion affects tiny object detection are lacking in modeling studies. Simply put, the role of visual motion, i.e., a conspicuous but underemphasized feature in existing models, in the detection of small objects remains poorly understood. We address this question by developing a parsimonious model that correlates the visual motion of the object’s leading edge with the luminance of its trailing edge. The results show that properly combining visual motion and luminance features can ensure the detection of small moving objects without the need for additional complex mechanisms. An evaluation of real-world datasets of small objects, including drones in infrared videos, indicates that the model outperforms existing methods inspired by flying insects and thus demonstrates potential for application in artificial vision systems.

Figures