A Deep Learning Framework for Accurate Vehicle Yaw Angle Estimation from a Monocular Camera Based on Part Arrangement

Abstract

An accurate object pose is essential to assess its state and predict its movements. In recent years, scholars have often predicted object poses by matching an image with a virtual 3D model or by regressing the six-degree-of-freedom pose of the target directly from the pixel data via deep learning methods. However, these approaches may ignore a fact that was proposed in the early days of computer vision research, i.e., that object parts are strongly represented in the object pose. In this study, we propose a novel and lightweight deep learning framework, YAEN (yaw angle estimation network), for accurate object yaw angle prediction from a monocular camera based on the arrangement of parts. YAEN uses an encoding–decoding structure for vehicle yaw angle prediction. The vehicle part arrangement information is extracted by the part-encoding network, and the yaw angle is extracted from vehicle part arrangement information by the yaw angle decoding network. Because vehicle part information is refined by the encoder, the decoding network structure is lightweight; the YAEN model has low hardware requirements and can reach a detection speed of 97FPS on a 2070s graphics cards. To improve the performance of our model, we used asymmetric convolution and SSE (sum of squared errors) loss functions of adding the sign. To verify the effectiveness of this model, we constructed an accurate yaw angle dataset under real-world conditions with two vehicles equipped with high-precision positioning devices. Experimental results prove that our method can achieve satisfactory prediction performance in scenarios in which vehicles do not obscure each other, with an average prediction error of less than 3.1° and an accuracy of 96.45% for prediction errors of less than 10° in real driving scenarios.

A Review of Generative Models for 3D Vehicle Wheel Generation and Synthesis

“…Using an alternative method, [98] uses a 3D Convolutional Neural Network (3D CNN) to identify pedestrians near autonomous cars. The way their architecture is set up, pedestrian entities are identified in the object detection components and then processed by the CNN for pedestrian identification and classification (Fault Interpretation Using Neural Networks, n.d.) [104]. One notable use is the YOLO v3 Convolutional Neural Network architecture, which is similar to other approaches in the field.…”

Prediction for Future Yaw Rate Values of Vehicles Using Long Short-Term Memory Network

“…Huang et al developed a monocular camera-based deep learning framework called YAEN (yaw angle estimation network). Unfortunately, their approach requires multiple times more hardware resources than is available for an electronic brake system [ 24 ]. Cunliang et al proposed a novel steering angle prediction YOLOv5-based end-to-end adaptive neural network control for vehicles.…”

Emma: An accurate, efficient, and multi-modality strategy for autonomous vehicle angle prediction

“…Specifically, Emma leverages both the images captured by the camera and the signals from inertial measurement unit (IMU) sensor with a fusion network to achieve feature fusion and multi-modal prediction. Moreover, we design and implement a few-shot learning module based on the most advanced meta-learning concept [6] to equip Emma with the ability of fast domain adaptation. That is, Emma can quickly adapt to various scenarios, such as new vehicle models and new road conditions.…”