Celestin Nkundineza

Dr. Celestin Nkundineza

Addis Ababa University; University of Nebraska- Lincoln

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 September 2024 in IEEE Access
Verified
Journal Article
Open access
JNJean Pierre Nyakuri
Jean Pierre Nyakuri
CNCelestin Nkundineza
Celestin Nkundineza
OGOmar Gatera
Omar Gatera
KNKizito Nkurikiyeyezu
Kizito Nkurikiyeyezu
Plant pest and disease management, especially in the early stages of infestation, is a critical challenge that poses significant threats and has potential to devastate agricultural crops, causing total yield loss and food insecurity. Traditional inspection methods are time-consuming and prone to errors due to limited labor expertise. Therefore, to tackle these challenges, harnessing advanced technologies such as artificial intelligence (AI), Machine Learning/Deep Learning (ML/DL), and Internet of Things (IoT) is essential for managing and mitigating agriculture hazards. This research presents a comprehensive review of the state-of-the-art DL architectures integrated with IoT-based systems applied to plant pest and disease detection (PPDD) by investigating different potential approaches that have been employed using DL and IoT up to the year 2024 to address challenges in agriculture. Convolutional Neural Network (CNN) architectures for image recognition, object detection, and their integration with IoT, embedded into mobile devices and unmanned aerial vehicles (UAV) are explored. Moreover, the research discusses the advantages and limitations of these techniques, emphasizing their architecture design, efficiency and accuracy. The findings demonstrate that there is a tradeoff between robustness and complexity among existing techniques, and authors recommend future trends aimed at creating robust models with fewer parameters that are more accurate and easily implementable on small IoT-based and portable devices suitable for in-field and real-time applications. Furthermore, while existing review papers discuss either DL or IoT separately, this research paper uniquely focuses on their combined models, providing a comprehensive overview of the synergistic potential of leveraging IoT-driven technologies alongside advanced DL algorithms to ease the task of researchers in the field of precision agriculture particularly in PPDD.

6 January 2020 in Sensors
Verified
Journal Article
Open access
PDF file
Pacifique TURABIMANAPacifique TURABIMANA
Pacifique TURABIMANA
CNCelestin Nkundineza
Celestin Nkundineza
The maintenance of railway systems is critical for their safe operation. However some landscape geographical features force the track line to have sharp curves with small radii. Sharp curves are known to be the main source of wheel flange wear. The reduction of wheel flange thickness to an extreme level increases the probability of train accidents. To avoid the unsafe operation of a rail vehicle, it is important to stay continuously up to date on the status of the wheel flange thickness dimensions by using precise and accurate measurement tools. The wheel wear measurement tools that are based on laser and vision technology are quite expensive to implement in railway lines of developing countries. Alternatively significant measurement errors can result from using imprecise measurement tools such as the hand tools, which are currently utilized by the railway companies such as Addis Ababa Light Rail Transit Service (AALRTS). Thus, the objective of this research is to propose and test a new measurement tool that uses an inductive displacement sensor. The proposed system works in both static and dynamic state of the railway vehicle and it is able to save the historical records of the wheel flange thickness for further analysis. The measurement system is fixed on the bogie frame. The fixture was designed using dimensions of the bogie and wheelset structure of the trains currently used by AALRTS. Laboratory experiments and computer simulations for of the electronic system were carried out to assess the feasibility of the data acquisition and analysis method. The noises and unwanted signals due to the dynamics of the system are filtered out from the sensor readings. The results show that the implementation of the proposed measurement system can accurately measure the wheel flange wear. Also, the faulty track section can be identified using the system recorded data and the operational control center data.

23 June 2025 in IEEE Access
Verified
Journal Article
Open access
JNJean Pierre Nyakuri
Jean Pierre Nyakuri
CNCelestin Nkundineza
Celestin Nkundineza
OGOmar Gatera
Omar Gatera
KNKizito Nkurikiyeyezu
Kizito Nkurikiyeyezu
Traditional Convolutional Neural Network (CNN) architectures face challenges in deployment on resource-constrained devices such as Internet of Things (IoT) platforms, mobile applications, and drones due to their computational intensity and memory requirements. This limitation motivates the development of lightweight models that maintain high performance with minimal resource consumption. This paper introduces Tiny MobileNetSE, a novel Hybrid CNN architecture designed for image classification tasks under critical resource constrained environments. This architecture integrates Squeeze-and-Excitation (SE) Blocks for adaptive feature recalibration, Batch Normalization (BN) for accelerated convergence, and applies knowledge distillation techniques from MobileNetV2 for enhanced feature generalization. We employed comprehensive experimentation to validate its performance across various metrics, including accuracy, number of parameters, and model size. Experimental results show that Tiny-MobileNet-SE achieved superior performance with a significantly reduced model size (0.72 MB) and a smaller number of parameters (482 K) compared to conventional lightweight CNNs, reaching up to 98.6% validation accuracy on real-world crop pest and disease datasets. This innovative architecture contributes to bridging the gap between computational efficiency and classification accuracy, enabling deep learning (DL) models to be deployed in small mobile, embedded, and IoT devices with minimized resource consumption.

9 March 2017 in Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit
Verified
Journal Article
Closed access
CNCelestin Nkundineza
Celestin Nkundineza
JTJoseph A Turner
Joseph A Turner
Railroad transportation is very important for the economic growth. The effective maintenance of railroad transportation is a critical factor for its economic sustainability. The high repetitive forces from a moving railcar induce cyclic stresses that lead to bending and potential deterioration of rails due to the initiation and propagation of fatigue cracks. Previous research on the prediction of fatigue life has been done under the assumptions of a uniform track bed and a homogeneous rail. However, the spatial variation of track stiffness is expected to increase the maximum stresses in the rails and, therefore, accelerate the fatigue process. This study is focused on the variations of the track modulus and the impact on fatigue life. The computational procedure is based on several hundreds of finite element models of the rails across a set of crossties chosen from a random ensemble with representative statistical variations. The mean of the track moduli is estimated from the field track deflection dynamic measurement data in comparison with the deflection data from the FE models. A multiaxial fatigue model is used for the estimation of fatigue cycles to crack initiation. The results show that a non-uniform track bed can reduce the fatigue life by up to 100 times in comparison with the behavior expected for a uniform track bed. The results of this study are expected to improve the effective maintenance and scheduling of rail inspection.

1 January 2024 in Open Engineering
Verified
Journal Article
Open access
PDF file
ZNZainah Namayanja
Zainah Namayanja
CNCelestin Nkundineza
Celestin Nkundineza
BZBehailu Mamo Zewdie
Behailu Mamo Zewdie
Flanges are critical parts of railway wheels that guide the wheels on the track and prevent them from derailing. However, curved tracks have a great influence on wheel flange wear, which leads to increased maintenance costs, passenger discomfort, poor ride quality, and increased noise and vibrations. This research aims to analyze the effect of curved tracks on wheel flange thickness reduction using the multi-body simulation software SIMPACK. The wheel flange wear was calculated to determine how much material is lost in terms of the area of material loss for different curve radii of 50, 100, and 200 m. The obtained results show that at curved track sections of smaller radius, for example, 50 m, the occurrence of wheel flange thickness reduction is higher than that of curved track sections with larger radii. Also, high speeds such as 40 km/h, especially at the track curved sections with small radii, lead to an increase in wheel flange thickness reduction compared to track curved sections with large radii. The results of wear indices obtained using SIMPACK and VAMPIR closely matched with a small percentage error difference of 1.94%. The matching of these results validated the effectiveness of the methods used in this research to calculate wear. Therefore, these conclusions will assist maintenance personnel in developing countermeasures to minimize wheel flange thickness reduction, ensure proper re-profiling, and reduce maintenance costs.

23 March 2015
Verified
Conference Paper
Closed access
PDF file
CNCelestin Nkundineza
Celestin Nkundineza
JTJoseph A. Turner
Joseph A. Turner
The bending of rail due to the repeated loading from railcar wheels is a known source of rail fatigue. If rail stresses are sufficiently high, they can initiate and propagate fatigue cracks after repeated cyclic loading such that they ultimately result in rail failure. Previous analyses of stresses from wheel loads have primarily focused on track beds for which the track stiffness is assumed uniform across a length of many cross-ties. In reality, however, spatial variations of track stiffness are known to exist and are affected by many factors such as the weather. These stiffness variations can lead to stresses that are locally higher than those predicted using models based on uniform average track stiffness alone. The work presented here is focused on the influence of spatial variations of track stiffness along the rail with respect to the maximum stresses generated. A computational model of a rail on a set of cross-ties with a statistically varying stiffness is used to study the maximum stresses generated when the track stiffness is not spatially uniform. The mean and standard deviation of the local track stiffness are varied and the maximum stresses at various positions within the rail are examined. This computational procedure is repeated for an ensemble of local track stiffness profiles to acquire the needed statistics of the corresponding stresses. These stresses are then related to crack initiation and the expected rate of crack propagation relative to the given the statistics of the track stiffness. This work is anticipated to have application for rail maintenance and the scheduling of rail defect inspections.