Tiny-YOLO object detection supplemented with geometrical data
Ivan Khokhlov, Egor Davydenko, Ilya Osokin, Ilya Ryakin, Azer Babaev, Vladimir Litvinenko, Roman Gorbachev
Abstract
We propose a method of improving detection precision (mAP) with the help of the prior knowledge about the scene geometry: we assume the scene to be a plane with objects placed on it. We focus our attention on autonomous robots, so given the robot's dimensions and the inclination angles of the camera, it is possible to predict the spatial scale for each pixel of the input frame. With slightly modified YOLOv3-tiny we demonstrate that the detection supplemented by the scale channel, further referred as S, outperforms standard RGB-based detection with small computational overhead.
Topics & Concepts
Artificial intelligenceComputer visionObject detectionFocus (optics)Computer sciencePixelScale (ratio)Object (grammar)Object-class detectionPattern recognition (psychology)Plane (geometry)Edge detectionImage (mathematics)Feature (linguistics)Feature detection (computer vision)Face detectionMotion detectionChange detectionDetection theoryCorner detectionOrientation (vector space)Robotics and Sensor-Based LocalizationAdvanced Image and Video Retrieval TechniquesAdvanced Neural Network Applications