In reality, you can easily read all kinds of printed words and everything around you. Perhaps you have never thought about how to deal with the visually impaired?
Statistics show that there are about 17 million visually impaired people in China, which means more than one in every 100 people are visually impaired. But we rarely see them in our daily life, because visually impaired people have a lot of inconvenience in their daily life, and it’s dangerous for them to go out. Could AI be their “eyes”?
Become Their eyes: The Story of Tracing Apps and the Visually impaired
If you could only have one App on your phone, which one would you choose? An Zhi, a visually impaired person, firmly gave the answer — plain description. Because the App allows him with poor eyesight to “see” the world in another way, words are no longer distant. He has “seen” beautiful poems, “heard” moving music, and even when he could not see the floor of the stairs alone, the line also accurately told him the answer, saved him from the dilemma. An App made him more brave to go out of the house and participate in a colorful life.
“The line is my eye.” Xiaojie, who is also visually impaired, has never been stingy in praising plain drawing, and even said plain drawing is a tool for visually impaired partners to survive. Line drawings help him solve life’s problems, from all kinds of electrical appliances, digital products, medicine instructions to all kinds of toiletries packaging can be recognized, he can also tell a picture book story to his cute little niece. It was something he could not have imagined without description.
How can the line drawing be treasured by the visually impaired community? Line drawing App is a simple and efficient OCR word recognition software. It is not only simple and easy to use, but also beautiful in design. It can easily realize a series of processes of “taking photos – identifying words – reading results”. At the same time, the line drawing is adapted to the mobile phone barrier-free auxiliary function, so the visually impaired can operate very easily on the mobile phone.
The original meaning of the word “plain drawing” is a literary writing technique. Lu Xun once summarized this technique into twelve words, that is, “there is real meaning, to whitewash, less affectation, do not show off”. Tao Xinle, the developer of the Line Drawing App, is just such a person. He observes the actual needs of different people and uses the code of the virtual world to meet the needs of people in the real world, making their lives better and more convenient.
Outline the journey of App developers: Cut into different scenarios and optimize product details
It might be hard to imagine that tao Xinle, a solo developer, first developed an App for his girlfriend. Tao xinle’s girlfriend loves reading and often takes notes. In order to reduce the pressure of transcribing, she tried using various types of word recognition software available at the time, but she found that some software was cumbersome, some expensive, and some inaccurate… See girlfriend pain unceasingly, tao Xinle decided to personally do a good experience, good effect of OCR character recognition tool for girlfriend use, and quickly put into action. This may be a romance unique to developers.
However, the development process of an App is full of unknowns and challenges. Under the conditions of that time, developing a software with OCR function faced a great challenge: how to make the word recognition fast and accurate. So he surveyed the vendors on the market who offered the service and compared their products to try to identify images of different scenes and find the one that worked best. Therefore, he found that the OCR technology capability and use experience of Baidu Brain AI open platform are the best, especially the recognition accuracy is better than other manufacturers, so he chose Baidu OCR technology without hesitation in 2017, and has been using it ever since.
However, excellent underlying technology does not mean all, Baidu Brain OCR has provided nearly 60 technical capabilities, good technology also needs to be applied to match the scene to play a greater value.
Therefore, Tao Xinle first made a subdivision study of the use of line drawing scenes, such as: Students take PPT in class to extract text and take notes, employees scan paper contracts into electronic version and make PDF, convert paper forms into Excel electronic version, translate text on pictures, teachers take photos and identify test questions and process them again, lawyers use them to extract text on paper documents and so on. In particular, he pays attention to and researches the special needs of visually impaired people.
After considering the user’s usage scenarios, it is also the continuous polishing of the product. At that time, there would often be recognition errors when pictures were transferred to text. In order to make up for this problem, the Line Drawing APP would carry out some technical processing before recognition, such as how to ensure the clarity of image compression and minimize the size of the picture; How to detect the automatic clipping of blank lines in the long graph without clipping text; How to segment the article automatically to make it easier for readers to read. These refined product designs ensure the clarity of pictures and make text messages easier to identify. After the recognition, the proofreading function of line drawing can make the recognition result and the original picture displayed on the same interface, so that users can quickly find the place that needs to be modified and edit on this basis.
Relying on Baidu’s excellent deep learning algorithm and pre-training model based on massive quality data, as well as the image preprocessing capability of the Tracing App, the key field recognition accuracy rate is 99%+. Seeing the smile on his girlfriend’s face when she used the line description made it all worthwhile for Tao, and he hoped more people could enjoy the pleasure.
Behind the success: with the heart of “craftsman” to carve the light of products
The tao Xinle of programmer one’s previous experience is doing a product all the time to carry on this matter “craftsman” heart. Tao xinle mentioned that the AI had a lot of difficulties in landing, and often walked forward while stepping on pits. When you encounter an unsolvable problem, you need to keep learning to overcome it.
Baidu brain OCR technology and countless developers like Tao Xinle side by side. As one of the earliest large-scale AI technologies, OCR technology continues to make breakthroughs in industrial applications. Baidu Brain OCR technology can provide multi-scene, multi-lingual and high-precision text detection and recognition services, with a number of ICDAR indicators ranking first in the world. It has been widely applied to remote identity authentication, fiscal and tax reimbursement, document electronization and other scenarios, reducing costs and increasing efficiency for enterprises, and bringing users more intelligent application experience.
, of course, the application of AI technology, in addition to the need of baidu brain providing leading ability of AI technology platform, also need more developers, such as pottery nahuy play imagination to apply AI in more real scene, meet different user groups, and even easier to ignore the needs of the disabled people, let the society more “AI”. At the same time, in order to reduce the threshold for independent developers and enterprises to independently train OCR character recognition models, Baidu Brain launched the industry’s first EasyDL OCR self-training platform, providing zero-threshold, customized, low-cost one-stop OCR model training service. This ensures high accuracy, meets the requirements of diversified scenarios, and effectively ensures data security.
In this era of science and technology empowering the public life, product design is an output of the concept of universal benefits. With over 8 million users, the App has become a word-of-mouth product in the industry. It is believed that in the future, more and more developers will create more intelligent applications combined with scenes through AI technologies and services provided by Baidu Brain AI Open platform, so that more people will live more convenient and better lives.
Immediately free experience Baidu OCR character recognition ability: ai.baidu.com/tech/ocr