Self-Supervised Object Distance Estimation Using a Monocular Camera

Hong, Liang; Ma, Zizhen; Zhang, Qian

doi:10.3390/s22082936

Cited by 16 publications

(6 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To alleviate inaccurate range estimations in different views, regression based on deep features can come in handy [62,[81][82][83][84][85]. Alike the FCN methods, the typical approach of deep regressors consists of two steps for object detection and distance regression.…”

Section: Range Estimation Approachesmentioning

confidence: 99%

See 1 more Smart Citation

Probabilistic 3D motion model for object tracking in aerial applications

Mirtajadini,

Amiri Atashgah,

Shahbazi

2024

IET Image Processing

View full text Add to dashboard Cite

Visual object tracking, crucial in aerial applications such as surveillance, cinematography, and chasing, faces challenges despite AI advancements. Current solutions lack full reliability, leading to common tracking failures in the presence of fast motions or long‐term occlusions of the subject. To tackle this issue, a 3D motion model is proposed that employs camera/vehicle states to locate a subject in the inertial coordinates. Next, a probability distribution is generated over future trajectories and they are sampled using a Monte Carlo technique to provide search regions that are fed into an online appearance learning process. This 3D motion model incorporates machine‐learning approaches for direct range estimation from monocular images. The model adapts computationally by adjusting search areas based on tracking confidence. It is integrated into DiMP, an online and deep learning‐based appearance model. The resulting tracker is evaluated on the VIOT dataset with sequences of both images and camera states, achieving a 68.9% tracking precision compared to DiMP's 49.7%. This approach demonstrates increased tracking duration, improved recovery after occlusions, and faster motions. Additionally, this strategy outperforms random searches by about 3.0%.

show abstract

Section: Range Estimation Approachesmentioning

confidence: 99%

“…More recently, refs. [81,83] replaced R-CNN with YOLO and added a distance regression module exploiting extracted deep features. While the previous methods were limited to a specific set of object classes that are detectable by object detectors, ref.…”

Section: Range Estimation Approachesmentioning

confidence: 99%

Probabilistic 3D motion model for object tracking in aerial applications

Mirtajadini,

Amiri Atashgah,

Shahbazi

2024

IET Image Processing

View full text Add to dashboard Cite

show abstract

“…For instance, [1] demonstrated a smartphone-based system for object detection, which, while effective, necessitates handheld operation and lacks distance estimation capabilities. Similarly, [2] explored the use of ultrasonic sensors attached to canes, offering some distance information but at the expense of user convenience and range of detection. Contrasting these approaches, our project innovates by integrating object detection and distance estimation into a single, wearable device, combining the convenience of a hands-free experience with the accuracy and immediacy of real-time feedback.…”

Section: B Project Statementmentioning

confidence: 99%

Obstacle Detection and Distance Estimation for Visually Impaired People

Leong,

Ramasamy

2023

IEEE Access

View full text Add to dashboard Cite

In the realm of assistive technologies for visually impaired persons (VIPs), existing solutions such as white canes and guide dogs have limitations in range and practicality. Moreover, current electronic systems often fall short in terms of portability and the ability to estimate distances in real-time. To bridge these gaps, this study introduces a revolutionary wearable device comprising a Raspberry Pi, a camera module, and a pretrained convolutional neural network, all integrated into a pair of smart glasses. These glasses are designed to identify objects and estimate their distances from the wearer, providing real-time auditory or haptic feedback. The development process was rigorous, involving the deployment of machine learning algorithms for object identification and the integration of camera and sensor technology into a lightweight, user-friendly frame. The system's performance was extensively evaluated using quantitative metrics, showing its precision, speed, and usability. Conclusively, this study presents a significant leap in wearable assistive technologies, offering enhanced spatial awareness, autonomy, and quality of life for VIPs.

show abstract

“…Compared to conventional stereo-based methods, self-supervised depth estimators provide widespread availability of training sequences of monocular video, but they need to simultaneously calculate depth and ego-motion by minimizing the photometric reprojection loss [51,56]. For the purposes of resolving the unknown scale factor through sequence input, a self-supervised learning network with a new reconstruction loss function was used with an average inference time of 12.5 ms per image [57]. As a faster state-of-the-art model, the Monodepth2 has acceptable results in the depth estimation used in an autonomous vehicle or smart navigation [58], which was enhanced from ResNet18.…”

Section: Introductionmentioning

confidence: 99%

Motorcycle Detection and Collision Warning Using Monocular Images from a Vehicle

Shabestari,

Hosseininaveh,

Remondino

2023

Remote Sensing

View full text Add to dashboard Cite

Motorcycle detection and collision warning are essential features in advanced driver assistance systems (ADAS) to ensure road safety, especially in emergency situations. However, detecting motorcycles from videos captured from a car is challenging due to the varying shapes and appearances of motorcycles. In this paper, we propose an integrated and innovative remote sensing and artificial intelligence (AI) methodology for motorcycle detection and distance estimation based on visual data from a single camera installed in the back of a vehicle. Firstly, MD-TinyYOLOv4 is used for detecting motorcycles, refining the neural network through SPP (spatial pyramid pooling) feature extraction, Mish activation function, data augmentation techniques, and optimized anchor boxes for training. The proposed algorithm outperforms eight existing YOLO versions, achieving a precision of 81% at a speed of 240 fps. Secondly, a refined disparity map of each motorcycle’s bounding box is estimated by training a Monodepth2 with a bilateral filter for distance estimation. The proposed fusion model (motorcycle’s detection and distance from vehicle) is evaluated with depth stereo camera measurements, and the results show that 89% of warning scenes are correctly detected, with an alarm notification time of 0.022 s for each image. Outcomes indicate that the proposed integrated methodology provides an effective solution for ADAS, with promising results for real-world applications, and can be suitable for running on mobility services or embedded computing boards instead of the super expensive and powerful systems used in some high-tech unmanned vehicles.

show abstract

Self-Supervised Object Distance Estimation Using a Monocular Camera

Cited by 16 publications

References 62 publications

Probabilistic 3D motion model for object tracking in aerial applications

Probabilistic 3D motion model for object tracking in aerial applications

Obstacle Detection and Distance Estimation for Visually Impaired People

Motorcycle Detection and Collision Warning Using Monocular Images from a Vehicle

Contact Info

Product

Resources

About