Re: 90 female school tyrants developed sign language translator to protect the dignity of aphasic
two people reminiscent of the image recognition project of Feng Ru cup (Feng Ru cup), when the project won the first prize, perhaps using this technique to help the aphasia. Wang Nana recalls, "we feel that we have found a need that nobody has found, to be the first to realize it, and to make what we have learned is valuable."
then, both of them began to study the feasibility of the sign language translation, and began to look up the papers, do experiments, write code, with a cup first prize bonus of 50 thousand yuan as the start of capital action. The image recognition was rejected because it was unable to carry the camera at any time; the gloves were rejected because it was so fascinating that the method of capturing the EMG signals through the arms was finally determined. Because of the popularity of smart watches, aphasers are wearing arms.
products are iterated in order to protect the dignity of the aphasic. Huang Shuang remembers that they have done the activity of "one day's aphasia". In the convenience store, because of the need to understand the price, the aphasia can only communicate with the shop assistant by mobile phone, and the person who has a long time in the back is impatiently urging. "If others know that I am a aphasic, I may be considerate, but I will not complain if I do not know, and the more I will complain, the more anxious I am." Some aphasic people do not want others to call them deaf people. They write themselves as "dragon people". After
determines the direction of research and development, the key is to continuously input data and debug data to ensure the accuracy of sign language translation. In order to test the reliability of the model, only two people at first recorded the data of the sign language, and wore the armband all day long.
"when you really help them, they know that they are good for them, and they will cooperate very well." Wang Nana said that, in order to ensure the diversity of the data and really let the aphasic people like it, they later contacted the Deaf College of Tianjin University of Technology, collaborate on data collection and intelligent closed-loop development of the arm, and reached a cooperative intention with the Chaoyang District residual Federation in Beijing to provide advice and guidance for the service of "hand sound" products. Zhang Quan also became a team adviser.
now, "hand tone" covers 200 sign language actions. In order to ensure accuracy, each action is recorded 1000 times. After having its own gesture database, the team also built a 7 layer BP neural network (a multilayer feedforward network trained by the error back propagation algorithm, one of the most widely used neural network models) to train the data speed, and the accuracy of recognition has reached 95%.
until last August, there were only two people in Wang Nana and Huang Shuang in the team, and later others gradually joined. They have no offices, meetings are "guerrilla warfare", coffee shops, restaurants, classrooms have been expropriated, research and development are in their own laboratories, but Wang Nana said: "we have not wanted to give up a day." Later, some people in the group quit because of going abroad. Wang Nana continued to study in North China Insurance Research Institute, and Huang Shuangbao went to Tsinghua University. The two female school tyrants did not abandon their team due to their studies and continued to firmly attract talent research and development products.
female dominated Technology entrepreneurial teams are rare, and two people have run through. At first, Huang Shuang met different opinions and waited for both sides to calm down and return to the dormitory to communicate again. Wang Nana told her that team communication is wrong and open to each other. Now two people can debate a "life and death" for one thing, and can also take a hand to eat and drink after the agreement is reached. There are other technicians in the team. When everyone stays up to write the code, two girls will occasionally turn into "programmer encouragement" to cheer for the team.
these two girls, who were born in 1994 and 1995, laughed and said, "maybe because we are not going to society, the distress of buying a house and the pressure of the society, we just feel that the technology we have is valuable to the society and can really help some people and we have time, so do it!"
they have many ideas about the future. "Hand voice" can realize the function of small talk and long conversation. In the future, the sign language teaching, the sign language letter and the intelligent arm loop can be introduced to solve the problems of the lack of sign language teachers. In addition, this artificial intelligence technology can be applied to other aspects after maturity, for example, children who can teach piano can play the piano more accurately.
the "hand" team won the first prize and the best Popularity Award, known as the most "inclusive" project in the "extreme future" responsibility innovation challenge jointly hosted by Baidu and the United Nations Development Programme (UNDP) recently.