YOLO-GG: a slight object detection model for empty-dish recycling robot

Ge, Yifei; Yue, Xuebin; Meng, Lin

doi:10.1109/icamechs57222.2022.10003347

Cited by 6 publications

(2 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The training phase is concerned with establishing an algorithm model that is taught to the system through training data, while the implementation phase involves applying the model that has been learned by the system during the training phase to produce certain decisions using testing data. Machine Learning can be carried out through two approaches, namely Supervised learning and Unsupervised learning [11], [12].…”

Section: Methodsmentioning

confidence: 99%

The Design of People Follower Robot to Transport Harvests Using Yolo V5 Detection

Imron,

Satria,

Nuraini

2023

SemanTIK

View full text Add to dashboard Cite

Robots transporting harvested products is commonly use automatic navigation systems with sensors such as lidar, cameras and ultrasonic proximity sensors are important for detecting obstacles and avoiding collisions. Mapping and positioning algorithm are crucial for precise the robot determine position in the field. A dependable drive system, adaptable to diverse terrains, and a robust motor are essential components. Efficient energy requirements must be addressed by using long-lasting battery systems and, where possible, automatic charging solutions. Robots picking mechanisms that can adapt to different types of plants, as well as storage systems that protect vegetables from damage during transport, also need to be considered. This writing utilizes does not huge robot harvest hardware as a prototype due to research scope limitations. Instead, it involves instructing a Jetson Nano and Arduino Uno to direct the camera towards the tracked object using coding through python and Arduino IDE. Computer mini used Jetson Nano for running the best model object tracking. Output serial connected with Arduino for control actuator like motor DC and servo motor. The method used is YOLOv5, while the image processing technique employs object detection, which likely serves for object identification and localization. The additional computer vision facilitates the camera to adapt to the objects. Through testing, the system successfully detects objects, particularly vest, with an average confidence value of 0.91%, indicating that the detection system has a very high level of accuracy in identifying and detecting vest. Keywords; Robot, Jetson nano, Arduino Uno, Object Detection, Yolov5

show abstract

Section: Methodsmentioning

confidence: 99%

The Design of People Follower Robot to Transport Harvests Using Yolo V5 Detection

Imron,

Satria,

Nuraini

2023

SemanTIK

View full text Add to dashboard Cite

show abstract

“…Convolution neural networks (CNNs) have been proven to be effective in various applications [1][2][3]: object detection [4,5], cultural heritage protection [6], environment monitoring [7], robotics [8][9][10] and healthcare [11].…”

Section: Introductionmentioning

confidence: 99%

Deep Learning Architecture Improvement Based on Dynamic Pruning and Layer Fusion

Meng

2023

Electronics

Self Cite

View full text Add to dashboard Cite

The heavy workload of current deep learning architectures significantly impedes the application of deep learning, especially on resource-constrained devices. Pruning has provided a promising solution to compressing the bloated deep learning models by removing the redundancies of the networks. However, existing pruning methods mainly focus on compressing the superfluous channels without considering layer-level redundancies, which results in the channel-pruned models still suffering from serious redundancies. To mitigate this problem, we propose an effective compression algorithm for deep learning models that uses both the channel-level and layer-level compression techniques to optimize the enormous deep learning models. In detail, the channels are dynamically pruned first, and then the model is further optimized by fusing the redundant layers. Only a minor performance loss results. The experimental results show that the computations of ResNet-110 are reduced by 80.05%, yet the accuracy is only decreased by 0.72%. Forty-eight convolutional layers could be discarded from ResNet-110 with no loss of performance, which fully demonstrates the efficiency of the proposal.

show abstract