Today our lives revolve around our smartphones. For anything we need the first resource is our smartphone, be it directions using google maps or networking through social media or browsing or using an app to order groceries or book a cab. Everything is possible with our smartphones. Well keeping this in mind, Microsoft has made one more thing possible with the use of smartphones. Using your smartphone, now you can identify any flower you see with the help of Smart Flower Recognition System.
Flower lovers and botanists are welcoming this move which will now allow them to identify any flower in their vicinity. There are around 2,50,000 species of flowers and this technology will help to identify and differentiate the flowers. You just need to click a picture of the flower with your smartphone and send it to the Smart Flower Recognition System and you will get your answer.
Microsoft Research Asia has developed Smart Flower Recognition System in collaboration with botanists at the Institute of Botany, Chinese Academy of Sciences (IBCAS). This collaboration was not a planned one but was by chance. Yong Rui, assistant managing director of Microsoft Research Asia (MSRA), was presenting on the topic image-recognition technology and this prompted the botanists at IBCAS, whose efforts to collect data on regional flower distribution were experiencing poor results, to ask MSRA to support them . They realized that MSRA’s image-recognition technology could be of help and so both collaborated and the result was Smart Flower Recognition System.
IBCAS had accumulated 2.6 million images of flowers uploaded by the public. And as anyone could upload picture to this data set, it was necessary to develop a system to filter out the bad images as it was not possible for anyone to supervise the uploads personally. To carry out this, MSRA had to develop algorithms so that bad pictures could be filtered out. The final result was 8, 00,000 pictures were filtered and then it was paired with the deep-learning network Caffe.
Microsoft’s Research Blog explains the working of this technology- ‘During the forward pass, each filter is convolved across the width and height of the input volume, computing the dot product between the entries of the filter and the input. This produces a 2-dimensional activation map of that filter. As a result, the network learns filters that activate per specific types of features at a given spatial position in the input’
With the efforts put in by the researchers, they have managed to get the Smart Flower Recognition system almost at par with humans. So this technology now can recognize the type of flower it sees with 90+ percent accurate rate.
This project has been very useful to the Chinese botanists in meeting their goals. This technology of flower- recognition has helped the experts in this field to acquire plant distribution in China in an efficient way.
Going forward, Microsoft and IBCAS plan to continue their collaboration and utilize this technology to build it further to make more apps which will be targeted at general public and field experts.
Try deep learning using MATLAB