KEYWORDS: Education and training, Convolution, Neural networks, Detection and tracking algorithms, Data modeling, RGB color model, Performance modeling, Feature extraction, Deep learning, Neurons
In recent years, research on distracted driving behavior recognition has made significant progress, with an increasing number of researchers focusing on deep-learning-based algorithms. Aiming at the problems of the existing distracted driving recognition algorithm, such as its oversized model and difficulty in adapting to low computing environments, a lightweight network MobileNetV2, is chosen as the backbone network and improved to design a distracted driving behavior detection method that is both accurate and practical. The Ghost module is employed to replace point-by-point convolution to reduce the computation, the Leaky ReLU function helps mitigate the problem of dead neurons, as it prevents gradients from becoming zero for negative inputs. Finally, the channel pruning algorithm is used to further reduce the model parameters. The experiment results on the State Farm dataset show that the model’s test accuracy can reach 94.66%, and the number of parameters is only 0.23 M. The improved model has significantly fewer parameters than the baseline model, which demonstrates the effectiveness and applicability of the method.
In order to better solve the social hot issue of garbage classification difficulties, we designed an intelligent garbage classification system based on the lightweight convolutional neural network MobileNet V2. The system can automatically realize the intelligent classification operation of garbage, and liberate people from the tedious and mechanical garbage classification. The system uses the Pytorch deep learning framework to build the MobileNet V2 network and train the garbage classification model, so that the model can be mounted on the Raspberry Pi to achieve embedded integration. This design realizes the highly integrated, intelligent, and data visualization of the garbage classification system, with accurate classification It has the characteristics of high density, simple and convenient use, high practical value and good applicability, which can greatly simplify the process of garbage classification.
The dynamic pattern generator of resolution test target is widely used to evaluate the imaging
quality of the optical system. The dynamic pattern generator is composed of resolution test target and
linear motion device and it can provide uniform motion within 400 millimeters. The resolution test
target is photoengraved on the demands of photoelectric detection system and their patterns are
different to be replaced. The length of the resolution test targets are specially designed to reduce the
difficulty of the time system synchronism. The linear motion device is made of linear motor servo
system. It can implement high speed, high precision and reliable motion. The servo system is designed
with current loop, speed loop and position loop. The current loop and the speed loop are the inner
control loops of the servo system and they can compensate the influence of the power fluctuation and
nonlinear, uncertain factors to lessen the burden of the main loop. The inner loop employs proportion
and integration (PI) algorithm without derivation feedback. The position loop, which belongs to the
outer control loop, adopts proportion, integration and derivation (PID) algorithm and forward feedback
algorithm. The Hall sensors are feedbacks of the current loop, the grating rule is the feedback for both
the speed loop and position loop. The servo system is designed for the speed stability, so the speed is
the control goal. Especially, control of high speed and low speed is implemented by speed and
acceleration forward feedback. The servo system can get the precision of 1%.
This measuring machine is specially designed to solve many parameters of crankshaft for some factory. According to the laser slot scanning principle, non-contact measuring of two axes' parallel degree for crankshaft is realized. The machine use laser technique, photoelectric measuring technique, fine mechanical technique, modern photoelectric sensor technique, electronics technique and computer-technique. And the difficulty which include complex shape, large weight, large size, high request of the measured parameter precision and difficult measuring, is solved.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.