ICACT20210159 Question.1
Questioner: baniyab@gram.edu 2021-02-09 오후 4:35:44 |
ICACT20210159 Answer.1
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:35:44
Chrome  Click!! |
| Would you mind to list out the audio and video features? Did you try to select the discriminative features from feature pool? |
Audio and video features are automatically completed according to the platform we use. The focus of our work is to develop an algorithm for the mixed recognition of gesture and speech, that is, using the platform as a bridge to connect the technologies, effectively superposition our recognition technologies to achieve the purpose of mixed and efficient recognition |
ICACT20210159 Question.2
Questioner: patcharin.ka@kmitl.ac.th 2021-02-09 오후 4:19:51 |
ICACT20210159 Answer.2
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:19:51
Chrome Click!! |
| why do you think iFLYTEK will give a better result than Hmm-gmm? what is a special technical of iFLYTEK? |
Here we comprehensively consider gesture and speech recognition. Considering the problems of users and the connection of devices, we focus on the use of gesture recognition and speech recognition as an auxiliary to gesture recognition.IFlytek input method can constantly modify the voice model with the identified content. This data is massive. It is no problem to collect hundreds of millions of pieces in a very short time, which meets the requirements of project development. IFlytek is in a virtuous circle of "users -- using users' voice to improve the SR model -- more accurate model recognition and better effect -- more users join". |
ICACT20210159 Question.3
Questioner: patcharin.ka@kmitl.ac.th 2021-02-09 오후 4:25:36 |
ICACT20210159 Answer.3
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:25:36
Chrome Click!! |
| Could you please explain more about Data preprocessing on cloud? what are processes inside this step? |
The cloud data processing, we use is the Modelarts AI in huawei cloud platform development platform, through the creation of OBS barrels, upload data set, at the same time, we develop a good gesture recognition algorithm, model training, the development of gesture recognition skills, use the platform that skill can be deployed on HiLens kit equipments, connect the terminal equipment of gesture recognition can be realized and display |
ICACT20210159 Question.4
Questioner: alihussain.dream@gmail.com 2021-02-09 오후 4:26:59 |
ICACT20210159 Answer.4
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:26:59
Chrome Click!! |
| Can you explain about your data? like, how many subjects data you have used. |
The number of samples is 5000+ |
ICACT20210159 Question.5
Questioner: franklinlu888@outlook.com 2021-02-09 오후 4:04:40 |
ICACT20210159 Answer.5
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:04:40
Chrome Click!! |
| How many sampless of training data for the gesture recognition? |
The number of samples is 5000+ |
ICACT20210159 Question.6
Questioner: alihussain.dream@gmail.com 2021-02-09 오후 4:40:32 |
ICACT20210159 Answer.6
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 4:40:32
Chrome Click!! |
| Do you think your method is good from the previous research work? if you think your proposed method is good. Can you highlight it into two or three points? |
1. Our work is to develop a gesture recognition algorithm with a high recognition rate, and at the same time, we will also carry out speech recognition. Of course, the two technologies are not simply superimposed, considering the environmental noise, we mainly focus on gesture recognition.
2. Our project has a user base, and we have consulted professional doctors on the visual recognition equipment, and we have reached cooperation.
3. Data processing after identification is also a feature of our project |
ICACT20210159 Question.7
Questioner: yanyonghang@henu.edu.cn 2021-02-09 오후 5:16:10 |
ICACT20210159 Answer.7
Answer by Auhor Yiile@outlook.com 2021-02-09 오후 5:16:10
Chrome Click!! |
| May I ask if your dataset is only elementary school? Or does it include all age groups? |
Hello, our gesture recognition recognizes specific gestures, while speech recognition recognizes specific voice features, and extraction is applicable to all age groups |
ICACT20210159 Question.8
Questioner: shubhamkj2405@gmail.com 2021-02-12 오후 7:39:22 |
ICACT20210159 Answer.8
Answer by Auhor Yiile@outlook.com 2021-02-12 오후 7:39:22
Chrome Click!! |
| Dear Author, that is a really good paper. I was wondering what changes would be required to detect gestures in the real-time video? Also, would it work on low-quality webcam cameras as your algorithm recognizes gestures in bad illumination? |
The principle of real-time video detection and face recognition is similar, both of which are feature extraction after image preprocessing, each frame of the video is obtained, and each frame of the image is processed. In addition, there is an anti-jitter mechanism.The obtained image is properly processed so that its features can be clearly displayed in the image. The main operations include: light compensation, image grayscale, Gaussian smoothing, image contrast enhancement, equalization histogram, etc.When the light is insufficient, the gesture is difficult, and speech recognition needs to be used to supplement the deficiency. Therefore, the method of using the mixed recognition of gesture and speech is reliable. Of course, in the actual application of school physical examination, there is basically no problem that the light is so poor that the detection is difficult, but there are more noises. |
