Because of changeable daylight and weather, the natural scenes of road are complicated. And the garbage on roads comes in polytropic shapes and sizes, especially small targets such as leaves. To detect the location and types of road-garbage in different scenarios quickly and accurately, we propose a road-garbage detection solution based on YOLOv5. The solution is finally applied to the garbage sweeper, where different deployment strategies are required for different modules. Firstly, to reduce the loss of image information caused by overexposure and underexposure, we design automatic exposure algorithm to adjust camera parameters in time and use OpenMP to accelerate it. Secondly, YOLOv5 algorithm based on object detection is trained for recognition and TensorRT framework is used for deployment. Taking into account the computing speed and accuracy, we choose FP16 computational precision for YOLOv5’s inference acceleration. Lastly, a low-cost computing platform named Jetson AGX Xavier is selected and the algorithm is optimized in combination with the characteristics of the hardware platform. Multithreading is also used to accelerate in software architecture. The results show that the average detection accuracy for various types of road-garbage reaches 70.4% under four different scenarios of self-built datasets. And the area of the smallest object detected accounts for 0.2‰ of the total area of the image. The speed of this road-garbage detection solution can reach 5fps when processing the 12-bit image of 2432*1226 size on the low-cost AGX Xavier computing platform.
Owing to the lack of defect samples in industrial product quality inspection, trained segmentation model tends to overfit when applied online. To address this problem, we propose a defect sample simulation algorithm based on neural style transfer. The simulation algorithm requires only a small number of defect samples for training, and can efficiently generate simulation samples for next-step segmentation task. In our work, we introduce a masked histogram matching module to maintain color consistency of the generated area and the true defect. To preserve the texture consistency with the surrounding pixels, we take the fast style transfer algorithm to blend the generated area into the background. At the same time, we also use the histogram loss to further improve the quality of the generated image. Besides, we propose a novel structure of segment net to make it more suitable for defect segmentation task. We train the segment net with the real defect samples and the generated simulation samples separately on the button datasets. The results show that the F1 score of the model trained with only the generated simulation samples reaches 0.80, which is better than the real sample result.
Considering the complexity of the button surface texture and the variety of buttons and defects, we propose a fast visual method for button surface defect detection, based on convolutional neural network (CNN). CNN has the ability to extract the essential features by training, avoiding designing complex feature operators adapted to different kinds of buttons, textures and defects. Firstly, we obtain the normalized button region and then use HOG-SVM method to identify the front and back side of the button. Finally, a convolutional neural network is developed to recognize the defects. Aiming at detecting the subtle defects, we propose a network structure with multiple feature channels input. To deal with the defects of different scales, we take a strategy of multi-scale image block detection. The experimental results show that our method is valid for a variety of buttons and able to recognize all kinds of defects that have occurred, including dent, crack, stain, hole, wrong paint and uneven. The detection rate exceeds 96%, which is much better than traditional methods based on SVM and methods based on template match. Our method can reach the speed of 5 fps on DSP based smart camera with 600 MHz frequency.
We carry out teaching based on optoelectronic related course group, aiming at junior students majored in Optoelectronic Information Science and Engineering. " Optoelectronic System Course Project " is product-designing-oriented and lasts for a whole semester. It provides a chance for students to experience the whole process of product designing, and improve their abilities to search literature, proof schemes, design and implement their schemes. In teaching process, each project topic is carefully selected and repeatedly refined to guarantee the projects with the knowledge integrity, engineering meanings and enjoyment. Moreover, we set up a top team with professional and experienced teachers, and build up learning community. Meanwhile, the communication between students and teachers as well as the interaction among students are taken seriously in order to improve their team-work ability and communicational skills. Therefore, students are not only able to have a chance to review the knowledge hierarchy of optics, electronics, and computer sciences, but also are able to improve their engineering mindset and innovation consciousness.
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.