Browsing by Author "Lin, Chun-Hsien"

Now showing 1 - 8 of 8

A Contactless Sport Training Monitor Based on Facial Expression and Remote-PPG
(2017-01-01)
Wu, Bing-Fei
;
Lin, Chun-Hsien
;
Huang, Po-Wei
;
Lin, Tzu-Min
;
Chung, Meng-Liang
To successfully increase athletes' or exercisers' fitness and endurance, the factors of physiological signal, emotion, or the level of fatigue should be considered during the training program. Many clinical decision support systems can assist to monitor the exercisers by some wearable devices. And, the questionnaire should also be taken into account to produce a report. Such process is cumbersome, and the results are not objective. Furthermore, one may feel uncomfortable when wearing the devices during the training program. In this research, the Rating of Perceived Exertion (RPE) is expected to be estimated automatically without any wearable devices and questionnaires. A camera based heart rate detection algorithm and a fatigue expression feature extractor are fused to estimate the RPE value. The results show that our heart rate detection algorithm can be competitive to the wearable devices, and the trend of the detected heart rate is correlated to RPE. Moreover, the fatigue feature can help reduce the error of the estimation.
A New Criterion of Human Comfort Assessment for Wheelchair Robots by Q-Learning Based Accompanist Tracking Fuzzy Controller
(2016-12)
Wu, Bing-Fei
;
Chen, Po-Yen
;
Lin, Chun-Hsien
;
電控工程研究所
;
Institute of Electrical and Control Engineering
Wheelchair users often struggle to drive safely and accompanied by caregivers. Wheelchair robots can employ the autonomous functions to avoid obstacles and reduce the workload of the caregivers. To achieve the goal, the accompanist needs to be steadily recognized and tracked by the robots. By using Q-learning based accompanist tracking fuzzy controller, wheelchair robots can peacefully follow the accompanist. Meanwhile, it is important to make the people on wheelchair robots feel comfortable. Based on ISO 2631-1, the ride qualities are obtained by averaging the acceleration data in the frequency bands, and the critical thresholds utilizing to assess the ride comfort are also determined inside. To make the level of bumpiness be as multiple as possible, four kinds of pavements are chosen in the experiments. The results show that the feelings of the occupants on the wheelchair robot are quite different from the ride comfort defined in ISO 2631-1, so a new standard is proposed to assess the ride comfort in this study. The accuracy of the proposed standard is 90.67 %, which is higher than that of ISO 2631-1, 42.48 %. Furthermore, to the best of our knowledge, this paper is thought to be the first one to present the ISO 2631-1-based comfort criterion for wheelchair robots.
Adaptive Feature Mapping for Customizing Deep Learning Based Facial Expression Recognition Model
(2018-01-01)
Wu, Bing-Fei
;
Lin, Chun-Hsien
;
電控工程研究所
;
Institute of Electrical and Control Engineering
Automated facial expression recognition can greatly improve the human-machine interface. The machine can provide better and more personalized services when it knows the human's emotion. This kind of improvement is an important progress in this artificial intelligence era. Many deep learning approaches have been applied in recent years due to their outstanding recognition accuracy after training with large amounts of data. The performance is limited, however, by the specific environmental conditions and variations in different persons involved. Hence, this paper addresses the issue of how to customize the generic model without label information from the testing samples. Weighted Center Regression Adaptive Feature Mapping (W-CR-AFM) is mainly proposed to transform the feature distribution of testing samples into that of trained samples. By means of minimizing the error between each feature of testing sample and the center of the most relevant category, W-CR-AFM can bring the features of testing samples around the decision boundary to the centers of expression categories; therefore, their predicted labels can be corrected. When the model which is tuned by W-CR-AFM is tested on extended Cohn-Kanade (CK+), Radboud Faces database, and Amsterdam dynamic facial expression set, our approach can improve the recognition accuracy by about 3.01%, 0.49%, and 5.33%, respectively. Compared to the competing deep learning architectures with the same training data, our approach shows the better performance.
Adaptive Neural Predictive Control for Permanent Magnet Synchronous Motor Systems With Long Delay Time
(2019-01-01)
Wu, Bing-Fei
;
Lin, Chun-Hsien
;
電控工程研究所
;
Institute of Electrical and Control Engineering
Since the permanent magnet synchronous motor system in this research needs about 40 ms to finish a control cycle, such a long delay in time strongly causes the bad performance for the conventional controllers, especially for position control. To well control the speed and position, an adaptive neural predictive control is proposed. A two-layer recursive neural network is employed as a speed predictor, and an extended Kalman filter is utilized to tune the parameters of the predictor adaptively. Chaos optimization algorithm and Newton-Raphson optimization are combined to solve the problem of predictive control. As for the speed control, the proposed method shows better performance. The position control is designed based on the speed control. Due to the physical limitation of the plant, the steady state error is still large. Hence, a fuzzy compensator is applied. From the experiment, the error is reduced obviously.
Chaotic Newton-Raphson Optimization Based Predictive Control for Permanent Magnet Synchronous Motor Systems with Long-Delay
(2015-01-01)
Wu, Bing-Fei
;
Lin, Chun-Hsien
;
電機資訊學士班
;
Undergraduate Honors Program of Electrical Engineering and Computer Science
A Tent-map chaotic Newton-Raphson optimization based neural network predictive control (TCNR-NPC) is developed to apply to the long-delay permanent magnet synchronous motor (PMSM) system in this paper. Due to a nonlinear model utilized in the predictive controller, nonlinear optimization methods turn into an important issue. To overcome the shortcoming of the conventional nonlinear programming on the initial condition sensitivity and maintain the accuracy of optimal solution, chaos optimization algorithm (COA) and Newton-Raphson (NR) are combined. With the comparison of COA and NR based optimization methods, our approach, the Tent-map chaotic Newton-Raphson (TCNR) optimization, is easier to reach the global optimum; thus, it would be employed in neural network predictive control. It is found that TCNR-NPC has a better performance than those of GPC, modified GPC, adaptive extended PSO based NPC, and PSO based PI controllers in real experiments.
Contrastive Feature Learning and Class-Weighted Loss for Facial Action Unit Detection
(2019-01-01)
Wu, Bing-Fei
;
Wei, Yin-Tse
;
Wu, Bing-Jhang
;
Lin, Chun-Hsien
;
電控工程研究所
Facial action unit detection (FAUD) is aimed to detect subtle facial motions, known as action units (AUs), induced by the traction or towing of facial muscles. In recent years, FAUD has become an attractive task since these slight changes on face reveal some cues of emotions which can then be utilized to infer people's affective. Many prior works are developed based on physical features of AUs, such as regional occurrence and temporal continuity, but few of them concern the negative impact of individual differences, causing an unrobust result in variant subjects and environments. To deal with the problem, a contrastive feature learning method is proposed to make a convolutional neural network (CNN) learn to extract the contrastive feature which is the difference between the features of a neutral face image and an AU-occurred face image. In this way, the individual information will be mitigated so that it becomes easier to detect AUs according to the features. A great disparity between the number of positive and negative samples, named as data imbalance, makes a serious interference in most detection problems, including FAUD. A lower occurrence rate of an AU leads a more severe data imbalance problem, which results in a biased result for the AU detection. Therefore, the class-weighted loss is proposed to change the weight-ratio between the positive and negative samples so as to encourage the learning of the positive samples and ease that of negative samples. Two widely used databases, BP4D and DISFA, are adopted to be the benchmark of the performance testing. In the experiment, it shows that the proposed method performs well both in accuracy and speed comparing with the state-of-the-art approaches.
Fully Convolutional Network for Crowd Size Estimation by Density Map and Counting Regression
(2018-01-01)
Wu, Bing-Fei
;
Lin, Chun-Hsien
;
電機工程學系
;
Department of Electrical and Computer Engineering
To handle the customer distribution in the certain areas, crowd counting is necessary for such applications, which is a labor-intensive work for human. Therefore, an automatic crowd counting system is in great demand, but it is still a challenging problem since the human heads and bodies are usually highly overlapping in crowd images. In this paper, a counting-by-regression framework is employed. The human head is modeled as a Guassian distribution. With a crowd density map estimator, the head count can be obtained by integrating over the density map. Most existing approaches only apply density map regression for training a density map estimator, but it is hard to find a suitable training parameters to train a good one; actually, the head count is overestimated easily. To mitigate this problem, counting regression is combined with density map regression. A deeper and lighter fully convolutional network (FCN) is designed to be a crowd density map estimator. The input and output size of the FCN are the same. After training by the proposed method, our model is more competitive comparing with others. The parameter quantity of the model is the lowest, and it needs the least inference time.
Motion Resistant Image-Photoplethysmography Based on Spectral Peak Tracking Algorithm
(2018-01-01)
Wu, Bing-Fei
;
Huang, Po-Wei
;
Lin, Chun-Hsien
;
Chung, Meng-Liang
;
Tsou, Tsong-Yang
Medical fields have seen increasing attention being given to image based heart rate measurement in recent years. One of the major limitations is motion artifacts of subject's head. Although there have been many studies focusing on signal extraction using different parameters and models, the development of frequency domain analysis is emerging slowly and moving in many directions. In the field of contact photoplethysmography (PPG), recent studies employed the acceleration signals to assist their spectral peak tracking algorithms. Inspired by the development of contact PPG, we are proposing a motion resistant spectral peak tracking (MRSPT) framework which eliminates the motion artifacts by integrating facial motion signals. The effectiveness of MRSPT coupled with the optimal image-based PPG (iPPG) signal has been tested against the state-of-the-art spectral peak tracking algorithms, multi-channel spectral matrix decomposition (MC-SMD), and the maximum peak selection coupled with optimal iPPG signal (Optimal MPS). Compared with MC-SMD and Optimal MPS, MRSPT uplifts the success rate-10 (success rate-5), the probability in which the absolute error is below ten (five) beats per mins, from 54.7% (36.3%) with MC-SMD and 73.0% (61.3%) with Optimal MPS to 90.7% (75.7%) with MRSPT in motion scenarios where subject moves arbitrarily with different distance or lighting. MRSPT also enhances the success rate-10 (success rate-5) from 40.7% (26.3%) with MC-SMD and 57.4% (45.7%) with Optimal MPS to 73.4% (58.4%) with MRSPT in all seven motion conditions including driving and running. Averagely, the success rate-five of Optimal MRSPT surpass the success rate-10 of both Optimal MPS and MC-SMD.