Four deep learning frameworks consisting of Yolov5m and Yolov5m combined with ResNet50, ResNet-10... more Four deep learning frameworks consisting of Yolov5m and Yolov5m combined with ResNet50, ResNet-101, and EfficientNet-B0, respectively, are proposed for classifying tomato fruit on the vine into three categories: ripe, immature, and damaged. For a training dataset consisting of 4500 images and a training process with 200 epochs, a batch size of 128, and an image size of 224 × 224 pixels, the prediction accuracy for ripe and immature tomatoes is found to be 100% when combining Yolo5m with ResNet-101. Meanwhile, the prediction accuracy for damaged tomatoes is 94% when using Yolo5m with the Efficient-B0 model. The ResNet-50, EfficientNet-B0, Yolov5m, and ResNet-101 networks have testing accuracies of 98%, 98%, 97%, and 97%, respectively. Thus, all four frameworks have the potential for tomato fruit classification in automated tomato fruit harvesting applications in agriculture.
ASME 2008 First International Conference on Micro/Nanoscale Heat Transfer, Parts A and B, 2008
This study presents a diffuser micropump and characterizes its output flow rates, like the parabo... more This study presents a diffuser micropump and characterizes its output flow rates, like the parabola shape on the frequency domain and the effecting factors. First, equivalent circuit using fluid-electric analogy was built up; then, the flow rate analysis results were compared to experiment results to verify the applicability of the circuit simulation. The operation frequency was 800 Hz for both cases and the maximum flow rates were 0.078 and 0.075 μl/s for simulation and experiment result, respectively. The maximum flow rate difference was 3.7%. The circuit then was used to analyze the inertial effects of transferred fluid as well as system components to the output flow rates. This work also explains why the flow rate spectrum has the shape of parabola. The analysis results showed that without inertial effects, the micropump flow rates are linearly proportional to the operation frequency; otherwise it has parabola shape. The natural frequency of the actuator-membrane structure was r...
In this study, a full-scale prototype of a water-based printer’s paint viscosity control device w... more In this study, a full-scale prototype of a water-based printer’s paint viscosity control device was proposed and conducted. Specifically, a novel approach of low-cost paint viscosity determination by exploiting the optical sensing method was proposed and a full-scale prototype was developed to exhibit the proposed concept. The feasibility of the proposed device was demonstrated by measuring the accuracy, setpoint control deviation and estimating its development cost. An accuracy of 1.5 cSt and a standard deviation of 1 cSt were recorded by comparing the measured viscosity data with the Zahn cup measuring method. The viscosity control results at a setpoint value of 40 cSt gave a variation of 0.8 cSt. The convergence time from initial 45 cSt to 40 cSt viscosity was 10 minutes. The developed cost was excessively competitive and was only one-fifth of the comparable commercial products. Furthermore, the device had solid and lightweight features, making it straightforward for portable app...
Four deep learning frameworks consisting of Yolov5m and Yolov5m combined with ResNet50, ResNet-10... more Four deep learning frameworks consisting of Yolov5m and Yolov5m combined with ResNet50, ResNet-101, and EfficientNet-B0, respectively, are proposed for classifying tomato fruit on the vine into three categories: ripe, immature, and damaged. For a training dataset consisting of 4500 images and a training process with 200 epochs, a batch size of 128, and an image size of 224 × 224 pixels, the prediction accuracy for ripe and immature tomatoes is found to be 100% when combining Yolo5m with ResNet-101. Meanwhile, the prediction accuracy for damaged tomatoes is 94% when using Yolo5m with the Efficient-B0 model. The ResNet-50, EfficientNet-B0, Yolov5m, and ResNet-101 networks have testing accuracies of 98%, 98%, 97%, and 97%, respectively. Thus, all four frameworks have the potential for tomato fruit classification in automated tomato fruit harvesting applications in agriculture.
ASME 2008 First International Conference on Micro/Nanoscale Heat Transfer, Parts A and B, 2008
This study presents a diffuser micropump and characterizes its output flow rates, like the parabo... more This study presents a diffuser micropump and characterizes its output flow rates, like the parabola shape on the frequency domain and the effecting factors. First, equivalent circuit using fluid-electric analogy was built up; then, the flow rate analysis results were compared to experiment results to verify the applicability of the circuit simulation. The operation frequency was 800 Hz for both cases and the maximum flow rates were 0.078 and 0.075 μl/s for simulation and experiment result, respectively. The maximum flow rate difference was 3.7%. The circuit then was used to analyze the inertial effects of transferred fluid as well as system components to the output flow rates. This work also explains why the flow rate spectrum has the shape of parabola. The analysis results showed that without inertial effects, the micropump flow rates are linearly proportional to the operation frequency; otherwise it has parabola shape. The natural frequency of the actuator-membrane structure was r...
In this study, a full-scale prototype of a water-based printer’s paint viscosity control device w... more In this study, a full-scale prototype of a water-based printer’s paint viscosity control device was proposed and conducted. Specifically, a novel approach of low-cost paint viscosity determination by exploiting the optical sensing method was proposed and a full-scale prototype was developed to exhibit the proposed concept. The feasibility of the proposed device was demonstrated by measuring the accuracy, setpoint control deviation and estimating its development cost. An accuracy of 1.5 cSt and a standard deviation of 1 cSt were recorded by comparing the measured viscosity data with the Zahn cup measuring method. The viscosity control results at a setpoint value of 40 cSt gave a variation of 0.8 cSt. The convergence time from initial 45 cSt to 40 cSt viscosity was 10 minutes. The developed cost was excessively competitive and was only one-fifth of the comparable commercial products. Furthermore, the device had solid and lightweight features, making it straightforward for portable app...
Uploads
Papers by Ngoc Bich Le