Sprint 3:oxford Project API tries to

This sprint we are mainly in the research and trial phase, mainly on the photo experience in the Voice interface part of the corresponding research and analysis.

Work Progress:

1. Image interface Design The work of the Zhao Yang and suggestion further, except for the corresponding interface introduced before, has the function of query and thumbnail suggestion, further improve the overall layout and design of the interface.

2. Oxford Voice Interface, the building investigates the use of the interface, and the sample code is analyzed in detail. and discuss it with us and give it in the appropriate document form.

3. Natural language processing NLP is an important part of our project, and audio to sentence is the work of the speech API, while sentence to query is the work of NLP. This part is the responsibility of Minlone, and has implemented the corresponding interface for initial debugging.

4. In terms of algorithm integration, Yandong has already set up the basic pipeline to successfully extract CNN feature. Whisk Foucs in the other feature integration work, to achieve a better overall algorithm flow.

This time we have established the specific work of the next Sprint4, but also for the Alpha release to do the final work:

1. Oxford API debugging work, this is responsible for Minlone and building.

2. NLP processing and the extraction of query are the responsibility of Minlone and Zhao Yang.

3. In the query to vector work needs to use Word vector thesaurus, try to use less vocabulary but the more general model of the mobile phone client to complete the transplant, this is responsible for whisk and building.

4. Complete the work of backstage service, this item is in charge of Zhao Yang.

5. Improve the accuracy of search, try to use other algorithms when vector word distance, and improve the algorithm of Multi-label image search.

Look forward to completing alpha release by the end of next week.

Sprint 3:oxford Project API tries to