Real-time Object Research Articles

Deteksi Otomatis Terhadap Pelanggaran Pembuang Sampah Menggunakan Metode You Only Look Once (YOLO)

Disposing of garbage is a bad thing that can spoil the view, cause bad smells, cause low to high level flooding, cause various diseases and can pollute the environment. Even though the ban on disposing of trash has been implemented, there are still many who violate it. The importance of avoiding this makes a study aimed at automatically detecting violations of waste disposal. The method used is YOLOv5, this method is an algorithm that can identify objects with high accuracy, besides that it can also carry out tracking processes in the form of bounding boxes for objects in real time. The programming language used is Google Colaboratory. The dataset used is in the form of 800 images and 2 videos. After testing the results of the research using the You only look once (YOLO) method, the best results were obtained on the parameter batch 5 epochs 5 with an accuracy of 95%. From these results it can be concluded that the use of the YOLO method is very accurate when applied to the detection process of an object.

Edge Real-Time Object Detection and DPU-Based Hardware Implementation for Optical Remote Sensing Images

The accuracy of current deep learning algorithms has certainly increased. However, deploying deep learning networks on edge devices with limited resources is challenging due to their inherent depth and high parameter count. Here, we proposed an improved YOLO model based on an attention mechanism and receptive field (RFA-YOLO) model, applying the MobileNeXt network as the backbone to reduce parameters and complexity, adopting the Receptive Field Block (RFB) and Efficient Channel Attention (ECA) modules to improve the detection accuracy of multi-scale and small objects. Meanwhile, an FPGA-based model deployment solution was proposed to implement parallel acceleration and low-power deployment of the detection algorithm model, which achieved real-time object detection for optical remote sensing images. We implement the proposed DPU and Vitis AI-based object detection algorithms with FPGA deployment to achieve low power consumption and real-time performance requirements. Experimental results on DIOR dataset demonstrate the effectiveness and superiority of our RFA-YOLO model for object detection algorithms. Moreover, to evaluate the performance of the proposed hardware implementation, it was implemented on a Xilinx ZCU104 board. Results of the experiments for hardware and software simulation show that our DPU-based hardware implementation are more power efficient than central processing units (CPUs) and graphics processing units (GPUs), and have the potential to be applied to onboard processing systems with limited resources and power consumption.

An Improved Lightweight Network Using Attentive Feature Aggregation for Object Detection in Autonomous Driving

Object detection, a more advanced application of computer vision than image classification, utilizes deep neural networks to predict objects in an input image and determine their locations through bounding boxes. The field of artificial intelligence has increasingly focused on the demands of autonomous driving, which require both high accuracy and fast inference speeds. This research paper aims to address this demand by introducing an efficient lightweight network for object detection specifically designed for self-driving vehicles. The proposed network, named MobDet3, incorporates a modified MobileNetV3 as its backbone, leveraging its lightweight convolutional neural network algorithm to extract and aggregate image features. Furthermore, the network integrates altered techniques in computer vision and adjusts to the most recent iteration of the PyTorch framework. The MobDet3 network enhances not only object positioning ability but also the reusability of feature maps across different scales. Extensive evaluations were conducted to assess the effectiveness of the proposed network, utilizing an autonomous driving dataset, as well as large-scale everyday human and object datasets. These evaluations were performed on NXP BlueBox 2.0, an advanced edge development platform designed for autonomous vehicles. The results demonstrate that the proposed lightweight object detection network achieves a mean precision of up to 58.30% on the BDD100K dataset and a high inference speed of up to 88.92 frames per second on NXP BlueBox 2.0, making it well-suited for real-time object detection in autonomous driving applications.

Dual-band real-time object identification via polarization reversal based on 2D GeSe image sensor

Dual-band real-time object identification via polarization reversal based on 2D GeSe image sensor

Convolutional Neural Network-Based Robot Control for an Eye-in-Hand Camera

In past decades, much progress has been obtained in vision-based robot control theories with traditional image processing methods. With the advances in deep-learning-based methods, convolutional neural network (CNN) has now replaced the traditional image processing methods for object detection and recognition. However, it is not clear how the CNN-based methods can be integrated into robot control theories in a stable and predictable manner for object detection and tracking, especially when the aspect ratio of the object is unknown and also varies during manipulation. In this article, we develop a vision-based control method for robots with an eye-in-hand configuration, which can be directly integrated with existing CNN-based object detectors. The task variables are generated based on parameters of the bounding box from the output of any real-time CNN object detector such as you only look once (Yolo). To address the chattering problem of bounding box, long short-term memory (LSTM) is used to provide smoothed bounding box information. A vision-based controller is then proposed following task-space motion control design formulation in order to keep the object of unknown aspect ratio in the center of field of view of the camera. The stability of the overall closed-loop control system is analyzed rigorously using the Lyapunov-like approach. Experimental results are presented to illustrate the performance of the proposed CNN-based robot controller.

EdgeDuet: Tiling Small Object Detection for Edge Assisted Autonomous Mobile Vision

Accurate, real-time object detection on resource-constrained devices enables autonomous mobile vision applications such as traffic surveillance, situational awareness, and safety inspection, where it is crucial to detect both small and large objects in crowded scenes. Prior studies either perform object detection locally on-board or offload the task to the edge/cloud. Local object detection yields low accuracy on small objects since it operates on low-resolution videos to fit in mobile memory. Offloaded object detection incurs high latency due to uploading high-resolution videos to the edge/cloud. Rather than either pure local processing or offloading, we propose to detect large objects locally while offloading small object detection to the edge. The key challenge is to reduce the latency of small object detection. Accordingly, we develop, the first edge-device collaborative framework for enhancing small object detection with tile-level parallelism. It optimizes the offloaded detection pipeline in tiles rather than the entire frame for high accuracy and low latency. Evaluations on drone vision datasets under LTE, WiFi 2.4GHz, WiFi 5GHz show that outperforms local object detection in small object detection accuracy by 233.0%. It also improves the detection accuracy by 44.7% and latency by 34.2% over the state-of-the-art offloading schemes.

BiTNet: A lightweight object detection network for real-time classroom behavior recognition with transformer and bi-directional pyramid network

In this paper, we propose BiTNet, a real-time object detection network, for enhancing teaching quality and providing feedback through real-time analysis of student behavior in classroom. BiTNet addresses challenges faced by current methods, such as occlusion and small objects in images. Specifically, it incorporates an Efficient Transformer Block (ETB) and Efficient Convolution Aggregation Block (ECAB) for extracting image features. ETB uses convolutional multi-head self-attention (CMHSA) to capture contextual information thus improves the capability to recoginze occluded objects. Its homogeneous multi-branch design reduces the computational costs. The single aggregation operation of ECAB aggregates the extracted features into a single feature map, improving the accuracy of recognition while reducing the computational cost. Then, the semantic and positional information extracted from the backbone are fused in Bi-directional Feature Pyramid Network (BiFPN). BiTNet also employs a smaller size detection head for recognition of small objects. Experimental results on classroom behavior dataset demonstrate that our proposed method has impressive accuracy (82.9%) and speed (256.7 FPS on GPU, 30.7 FPS on CPU, and 18.9 on camera). We further experiment with our method on Pascal VOC 2007 and VisDrone2021, validating BiTNet’s generalization and its capability to recognize small objects. In conclusion, BiTNet has a simple yet efficient network structure and feature extraction module, which reduce computational cost while maintaining recognition accuracy, making it a contender for real-time classroom behavior recognition tasks.

Real-time tool detection in smart manufacturing using You-Only-Look-Once (YOLO)v5

Computer vision plays an essential role in Industry 4.0 by enabling machinery to perceive, analyze, and control production processes. Object detection, a computer vision technique that accurately classifies and localizes objects within images, has gained significant interest. This technique can be applied in various domains, including manufacturing, to assist in the detection of different tools. In this paper, You-Only-Look-Once (YOLO)v5 real-time object detection technique has been developed and optimized, to detect different tool types and their locations in a manufacturing setting. To train the neural network, a dataset of 3,286 tool images from the internet has been collected and annotated. To enhance the model's ability in generalization, three augmented variants of each image have been created to improve rotation invariance. The model's training scheme has been further optimized with stochastic gradient descent after configuring different hyperparameters such as learning rate and momentum. The fine-tuned model achieved a mean average accuracy of 98.3 %, demonstrating the high precision of the model in detecting different tool types and their locations in real-time.

Real-Time 3D Object Detection, Recognition and Presentation Using a Mobile Device for Assistive Navigation

Real-Time 3D Object Detection, Recognition and Presentation Using a Mobile Device for Assistive Navigation

Virtual Aquascape for Users Utilizing Augmented Reality

Aquascape is the art of arranging plants in water. The large number of complaints from aquascape buyers were not up to expectations, causing sellers to have difficulty in setting an example of an orderable aquascape, so researches created an aquascape application system to increase their promotional activities with augmented reality technology accessible through smartphone. One uses the methode MDLC (Multimedia Development Life Cycle). As a result, the application can implement a markless augmented reality to display 3D objects in real time with the help of a camera on a smartphone or a trilib feature by importing 3D object files found on file managers on a smartphone. User can easily provide the desired aquascape example of the aquascape application.

Multi-sensor fusion for real-time object tracking

Multi-sensor fusion for real-time object tracking

Digital Book of Indonesian Rare Flora Using Augmented Reality

The uniformity of the flora is still not widely known. Therefore, natural knowledge has an important role in education considering the limitations of print media as a learning medium. Innovation by utilizing Augmented Realityp technology is very helpful as a learning medium that can support the learning process more effectively and efficiently, on the other hand Augmented Reality that displays 3D models of objects in real time can also increase interest in learning in knowing Indonesia's rare flora attractively and interactively.So, researchers designed and developed the application "Digital Flora Langka Indonesia Using Augmented Reality" with SDLC (Software Development Life Cycle) methods which are expected to help as an interesting learning medium.

Research and Explainable Analysis of a Real-Time Passion Fruit Detection Model Based on FSOne-YOLOv7

Real-time object detection plays an indispensable role in facilitating the intelligent harvesting process of passion fruit. Accordingly, this paper proposes an FSOne-YOLOv7 model designed to facilitate the real-time detection of passion fruit. The model addresses the challenges arising from the diverse appearance characteristics of passion fruit in complex growth environments. An enhanced version of the YOLOv7 architecture serves as the foundation for the FSOne-YOLOv7 model, with ShuffleOne serving as the novel backbone network and slim-neck operating as the neck network. These architectural modifications significantly enhance the capabilities of feature extraction and fusion, thus leading to improved detection speed. By utilizing the explainable gradient-weighted class activation mapping technique, the output features of FSOne-YOLOv7 exhibit a higher level of concentration and precision in the detection of passion fruit compared to YOLOv7. As a result, the proposed model achieves more accurate, fast, and computationally efficient passion fruit detection. The experimental results demonstrate that FSOne-YOLOv7 outperforms the original YOLOv7, exhibiting a 4.6% increase in precision (P) and a 4.85% increase in mean average precision (mAP). Additionally, it reduces the parameter count by approximately 62.7% and enhances real-time detection speed by 35.7%. When compared to Faster-RCNN and SSD, the proposed model exhibits a 10% and 4.4% increase in mAP, respectively, while achieving approximately 2.6 times and 1.5 times faster real-time detection speeds, respectively. This model proves to be particularly suitable for scenarios characterized by limited memory and computing capabilities where high accuracy is crucial. Moreover, it serves as a valuable technical reference for passion fruit detection applications on mobile or embedded devices and offers insightful guidance for real-time detection research involving similar fruits.

Low Cost PID Controller for Student Digital Control Laboratory Based on Arduino or STM32 Modules

In the teaching process, it is important that students do not carry out exercises only by computer simulations, but also that they carry out research in real time. In times of distance learning during the COVID-19 pandemic, it would be necessary to find a solution so that the students can perform such exercises individually at home. Therefore, it has become necessary to develop cheap and simple modules of digital controllers along with analog objects with adjustable order and time constants. This paper describes a low-cost proportional–integral–derivative (PID) controller for teaching students control techniques and analog control objects in real time. The PID controller is based on the cheap and widely available microcontroller modules Arduino or STM32. The advantage of this solution is that the algorithm of the digital PID controller is calculated every constant period of time. Both the solutions presented in the paper have been successfully tested by students in practice during remote learning during the COVID-19 pandemic.

CNN-Based Vehicle Bottom Face Quadrilateral Detection Using Surveillance Cameras for Intelligent Transportation Systems.

In intelligent transportation systems, it is essential to estimate the vehicle position accurately. To this end, it is preferred to detect vehicles as a bottom face quadrilateral (BFQ) rather than an axis-aligned bounding box. Although there have been some methods for detecting the vehicle BFQ using vehicle-mounted cameras, few studies have been conducted using surveillance cameras. Therefore, this paper conducts a comparative study on various approaches for detecting the vehicle BFQ in surveillance camera environments. Three approaches were selected for comparison, including corner-based, position/size/angle-based, and line-based. For comparison, this paper suggests a way to implement the vehicle BFQ detectors by simply adding extra heads to one of the most widely used real-time object detectors, YOLO. In experiments, it was shown that the vehicle BFQ can be adequately detected by using the suggested implementation, and the three approaches were quantitatively evaluated, compared, and analyzed.

Data Exchange Mechanism for Real-Time Object Detection in Cloud-Edge IoT System

Real-time target detection based on massive surveillance video data plays a key role in the smart grid, especially in power vision applications. With the indepth construction of power Internet of Things, centralized cloud applications pose challenges to real-time applications due to high bandwidth costs, resulting in delays and loads. Edge computing attempts to perform all the tasks on the edge. However, limited resources on the edge generally cannot guarantee real-time tasks, while complex intelligence algorithms are difficult to fully deploy. Given that both cloud computing solution and edge-only computing solution cannot achieve a good effect, this paper proposes a data exchange mechanism for real-time object detection in a cloud-edge IoT system, which achieves a balance of object detection task executing between delay and accuracy by utilizing the designed intelligent task scheduler. Experimental results performed in real environments using real surveillance video datasets show that by using our cloud-edge collaboration mechanism, the task response time of the system is only about 12% of cloud computing solution while ensuring query accuracy within the practical scope. The system also improves the query accuracy by about 59% compared to edge computing solutions, while the task response time was reduced to its 72%.

Real-Time Information Fusion System Implementation Based on ARM-Based FPGA

In this study, an information fusion system displayed fusion information on a transparent display by considering the relationships among the display, background exhibit, and user’s gaze direction. We used an ARM-based field-programmable gate array (FPGA) to perform virtual–real fusion of this system as well as evaluated the virtual–real fusion execution speed. The ARM-based FPGA used Intel® RealsenseTM D435i depth cameras to capture depth and color images of an observer and exhibit. The image data was received by the ARM side and fed to the FPGA side for real-time object detection. The FPGA accelerated the computation of the convolution neural networks to recognize observers and exhibits. In addition, a module performed by the FPGA was developed for rapid registration between the color and depth images. The module calculated the size and position of the information displayed on a transparent display according to the pixel coordinates and depth values of the human eye and exhibit. A personal computer with GPU RTX2060 performed information fusion in ~47 ms, whereas the ARM-based FPGA accomplished it in 25 ms. Thus, the fusion speed of the ARM-based FPGA was 1.8 times faster than on the computer.

Complementary spatial transformer network for real-time 3D object recognition

Complementary spatial transformer network for real-time 3D object recognition

Katalog Digital Untuk Promosi Tas Wanita Berbasis Augmented Reality

In today's era, marketing techniques use brochures, pamphlets, video advertisements, newspapers and word of mouth. This has a deficiency in consumer confidence because it cannot interact directly with the product of interest. A good marketing strategy methodology will influence buyer interest and increase sales traffic. In this case it provides opportunities for MSMEs, especially UD. Creative to increase sales traffic. Business progress can be seen from the marketing sector which continues to increase. MSME entrepreneurs use marketing methods in the form of pamphlets and word of mouth. This affects the lack of increased sales. This application is able to provide experience to potential customers to interact directly with products in Augmented Reality. This application displays 3D objects in real time and sees the shape of the object in detail, so that potential customers have more confidence in the product to be purchased. This application feature can add other objects stored in external storage or in the form of links, so users don't have to bother rebuilding the application when they want to add other objects

A memphotoristor for high-efficiency temporal vision processing

A photodetector with history-dependent dynamical responses to optical inputs could offer an essential breakthrough in performing temporal vision processing and time-series prediction without the requirement for sophisticated circuitry and undesired high energy consumption. Till now, memristor dynamics and its effective modulation with photo illumination have been utilized to mimic bio-inspired vision processing. Despite significant effort, it is still challenging to process real-time analogue temporal optical information with high accuracy, which requires developing a photonic counterpart of the electric-triggered memristor. Here, we report on the development of a Ga2O3-based proof-of-concept memphotoristor, in which memristive dynamics were stimulated with ultraviolet photon flux rather than conventional voltage. Specifically, a distinct hysteresis loop opening appeared in the cyclic photocurrent-ultraviolet intensity curves, where the magnitude of the loop opening depends on the photon flux. Additionally, light-illumination-induced fading memory, nonlinearity, and temporal dynamics were successfully utilized to demonstrate in-sensor reservoir computation and time-series prediction with an accuracy of 98.86%. Our research will pave the way to developing a wide range of cutting-edge optoelectronics for various applications, such as photosensors, photonic memory storage, and neuromorphic vision sensing of objects in real time.