Mobile QR Code QR CODE

Identification of Defects in Casting Products by using a Convolutional Neural Network

(Dilliraj Ekambaram) ; (Vijayakumar Ponnusamy)

The main perspective when ensuring dependability in speculations over accuracy in casting parts is a project quality confirmation process that is both careful and meticulous under Industry 4.0. When thorough and extensive casting project examination strategies merge with expanded metal project quality standards, casting production, augmented visual inspections, ensemble process modification and execution are improved. In this paper, we use publicly available casting image datasets for visual inspection, which classify defective and non-defective casting. Inspired by the convolutional neural network (CNN), we propose two-stage convolution for modeling, with DenseNet for classifying casting products. Through experimentation, we achieved an F1-score of 99.54% with a processing time of 454ms using a CPU for classification of casting product inspections. The modified modeling of the CNN in this work helps to improve optimization, compared to other basic machine learning mechanisms that measure quality.

Synthetic Image Generation for Data Augmentation to Train an Unconscious Person Detection Network in a UAV Environment

(Junghoon Sung) ; (Heegwang Kim) ; (Mingi Kim) ; (Yeongheon Mok) ; (Chanyeong Park) ; (Joonki Paik)

In this paper, we propose a data augmentation method using synthetic data generation for detecting an unconscious person in drone images. First, we extract the most salient and delicate foreground mask from a reference image that simulates an unconscious person situation using Net, which is a Salient Object Detection (SOD) model. Second, we apply shadow generation to the foreground mask for the natural appearance of the object. The unconscious person object generated by the foreground mask is synthesized with the background image of the Unmanned Aerial Vehicle (UAV) environment according to the altitude using object resizing. Therefore, we generate the most similar data to the image acquired by the drone. We verified the synthetic data-based image dataset using various object detection models, such as YOLOv4, YOLOv5, and EfficientDet. As a result, the Average Precision (AP) is higher than that of the real-world dataset. Our proposed method could be used to generate synthetic data for detecting an unconscious person and reducing the time cost and human resources needed for various tasks.

Design and Analysis of Approximate 4-2 Compressor for Efficient Multiplication

(Jahyung Gu) ; (Youngmin Kim)

Approximation computing concepts have been introduced and studied to provide sufficient accuracy with low power. Many approximate 4-2 compressor designs and methods are proposed to build approximate multipliers. However, most of the designs used for conventional approximate multipliers have been proposed to achieve low power by compromising accuracy or to maintain accuracy with low performance. In this study, a new 4-2 compressor is proposed to improve the error rate, delay, and power consumption. The proposed compressor was used to build an approximate multiplier. Transistor-level SPICE simulations with a high-performance 32-nm predictive technology model (PTM) were conducted to verify the efficiency of the proposed design. The results show that the proposed 4-2 compressor design can achieve up to 4.2% and 22.5% improvement in power consumption and delay, respectively. The proposed multiplier design can achieve up to 4.2% and 15.5% improvement in power consumption and error metrics, respectively. Compared to other designs, the proposed multiplier has lower error in small operand operations.

Continuous Blood Pressure Estimation using 1D Convolutional Neural Network and Attention Mechanism

(Youjung Seo) ; (Jungwhan Lee) ; (Unang Sunarya) ; (Kwangkee Lee) ; (Cheolsoo Park)

Patients with hypertensive blood pressure (BP) needs a round-the-clock BP monitoring and must take precautions to prevent emergencies such as stroke or heart failure. This paper suggests a deep neural network (DNNs?based BP estimation approach using electrocardiogram (ECG), photoplethysmogram (PPG), and ballistocardiogram (BCG) signals. The proposed approach consists of a one-dimensional convolutional neural network (1D CNN) followed by the attention mechanism known as Luong attention. Estimations under the proposed model yield mean absolute error (MAE) of 3.299±2.419 for systolic and 2.69±1.821 for diastolic BP. The algorithm can effectively predict BP without a recurrent neural network (RNNs), which is a typical DNNs model for processing sequential data. Additionally, the proposed approach is preferable owing to its ability to explain the model.

The Hardware Cost and Computing Accuracy Trade-off in Multipliers using Imprecise 4-2 Compressors for Image Processing Applications

(Yongqiang Zhang) ; (Cong He) ; (Xiaoyue Chen) ; (Guangjun Xie)

Approximate computing has been widely used in image processing applications to significantly reduce the hardware cost of circuits; however, this induces a sacrifice in computing accuracy. The compromise between accuracy and hardware cost in approximate multipliers has not been investigated yet. To address this issue, this paper proposes a set of approximate 8×8 Dadda multipliers built by using an efficient imprecise 4-2 compressor. The compressor introduces symmetrical errors into the truth table of the exact design to reach a simpler structure. Furthermore, as an important image processing application, image multiplication is implemented with the proposed multipliers. Synthesis and simulation results show that the overall performance of the multipliers varies depending on the various assessment criteria. Utilization of the modified compressor in the multipliers results in area, delay, and power reductions of 38%-72%, 14%-33%, and 39%-77%, respectively, compared to the exact design, while maintaining acceptable computing accuracy in image multiplication. According to the results, the proposed multipliers achieve a better trade-off between energy efficacy and computing accuracy than the existing designs, which could be certified as options for exact multipliers in image processing.

Dual-stage, Q-learning based, Shortest-path Routing Computing Scheme: Evaluated A Case Study on Illegal-parking Surveillance Networks

(Navin Ranjan) ; (Sovit Bhandari) ; (Yeong-Chan Kim) ; (Hoon Kim)

This paper explores a reinforcement learning (with Q-learning as the algorithm) for vehicle routing in a city-wide road network with a large customer (a generic name used for each of the locations (to denote a person possibly associated with this location) through which the vehicle needs to be routed) counts. Most reported research on vehicle routing had mainly focused on calculating the shortest route (while neglecting the shortest path between certain customer(s) and certain/all other customer(s)) between the customers. Hence, we proposed a dual-stage, Q-learning based, shortest-route generation (DQSR) in a road network, considering the shortest path between each customer and certain other customers. Notably, the first stage’s Q-learning agent develops the meta-graph of the shortest path between each customer and this customer’s nearest neighbor customers (chosen differently in number), instead of each other customer, and finds this path’s length. This development is motivated by the fact that any two adjacent customers in the shortest route are located closer than far apart, reducing the time complexity of the first stage’s Q-learning. Subsequently, the second stage’s Q-learning agent finds the shortest route from any customer to other customers from the meta-graph, connecting each other customer only once and returning to the starting customer. Hence, the DQSR generates the shortest route in a relatively shorter duration due to the small size of the state-action pairs in the meta-graph. Further, we conducted a case study on the DQSR-based, illegal-parking surveillance in Yeonsu-gu, Incheon, South Korea, to check the DQSR’s effectiveness. The case study demonstrates that the DQSR is more efficient than two other algorithms used for the same purpose in the case study, in terms of time complexity and shortest-route generation.

Intersection Management System for Autonomous Vehicles using a Fuzzy Inference System

(Muhammad Nadeem Ali) ; (Byung-Seo Kim)

An intersection management system (IMS) is a critical element in a transport system. Primarily intersection points are controlled with a time-based traffic signal. With the rapid growth in autonomous vehicles (AVs) and intelligent transport systems (ITSs), intersection management is an important area of research and development. The primary direction of this study is designing a management system that can communicate with vehicles, respond to them according to the traffic situation, and significantly reduce and avoid fatal situations like accidents. In this paper, a two-layered fuzzy logic-based controller is proposed to automate a cross-sectional intersection. Vehicles convey their particular information to the intersection unit, and the controller decides the specific action for the vehicles. This intersection controller is an IMS. After that, all roadside information is transferred to a centralized IMS (CIMS). The proposed CIMS reacts to a vehicle by giving an action command to leave the intersection. For prediction in a complex and dynamic environment, especially when many relevant factors overlap, fuzzy logic can be efficient. In this paper, fuzzy logic indicates the action for vehicles. Simulations were performed using MATLAB.

Multiband Slotted Crescent-shaped Patch Antenna for K-band Satellite and mmWave Communications

(Md. Najmul Hossain) ; (Liton Chandra Paul) ; (Md. Abdur Rahim) ; (Jungpil Shin)

In this paper, a multiband miniaturized crescent-shaped patch antenna with circular slots is presented for ultra-wideband applications. The proposed antenna is constructed on a Flame Retardant 4 (FR-4) dielectric substrate. Next-generation wireless communication systems, such as beyond 5G (B5G) and 6G mobile communication systems require a wider bandwidth and higher data rates. The overall size of the proposed antenna is 11×9×2.4 mm3 operating in the frequency range of 17.73 GHz to 26.04 GHz and 29.6 GHz to 31.02 GHz for satellite communications and radar systems, respectively, as well as 35.40 GHz to 38.65 GHz for future mmWave B5G mobile applications. These frequency ranges usually fall into the K-band and mmWave frequency spectrums. Simulation results demonstrate that the peak gain of the proposed antenna is 4.68 dB, which is obtained at 25 GHz.

An Optimal Smart Tank Juice-level Monitoring System for Beverage Industries: A Case Study of Raha Beverages Company Limited, Arusha, Tanzania

(Yvonne Iradukunda) ; (Elizabeth Mkoba) ; (Silas Mirau)

Poor monitoring of levels in juice tanks is among the challenges that beverage industries face when pumping liquid from one tank to another. This leads to spilling fluids, faulty juice tests, and industrial accidents. To keep track of the liquid level in a tank, various approaches have been used. Existing technologies are costly and not interactive, and the majority do not benefit individuals with physical disabilities when manual monitoring is needed. The purpose of this paper is to present an optimal smart tank juice-level monitoring system that can be used in beverage industries. The system is able to monitor the juice level within a tank and regulate a pump using voice commands via Alexa and the Amazon Echo Dot. The proposed system was tested and validated, with key findings being that the developed prototype prevented overflowing, accidents, and changes in juice flavor during the dilution process. This paper contributes to the body of knowledge for food and beverage industries in that engineers and operators of beverage industries can monitor the level of juice in a tank, as well as enhance communication when pumping juice from one tank to another in real time.

Comparative Analysis of Schnorr Digital Signature and ECDSA for Efficiency using Private Ethereum Network

(Jangho Na) ; (Hye-Young Kim) ; (Nohpill Park) ; (Beomjoo Seo)

This paper reports the use of the Schnorr Digital Signature as a new alternative instead of the Elliptic Curve Digital Signature Algorithm (ECDSA) currently used in Ethereum. Elliptic curve cryptography, which is the cryptographic background of blockchain digital signature, is explained in the background and related works. This study then examined, the problems facing ECDSA and how the Schnorr Digital Signature could solve them. The test was performed by comparing Schnorr and ECDSA implemented in a private Ethereum network environment. The performance results showed that Schnorr Digital Signature has strengths in terms of efficiency over ECDSA. The efficiency deals with the gas price, signature size, block size, block elapsed time, and signature algorithm calculation time.