Core demands
The FMCG industry has always used the traditional manual inspection method for the implementation audit of terminal stores. Due to the large number of stores nationwide, the cost of brand audit is also huge. Huihe Technology is committed to driving the digital marketing reform of FMCG industry with technology. For the display review of retail stores across the country, it uses strong AI image customization identification technology to solve the high cost and low efficiency problems of the current traditional way. Let retail stores complete display review by uploading display images independently, use AI technology to drive efficiency improvement, and establish a set of basic identification library for each retail store.
Solution
Collect image samples according to the requirements of display activities carried out by FMCG brands. Through customized image recognition models, multiple algorithm components and training templates provided by Baidu, train accurate models with a small amount of sample data based on Baidu big data, and quickly and accurately identify whether the target image is qualified.
Sample training: according to the requirements of brand display activities, positive and negative image samples were collected in the early stage, template training, online and other operations were carried out through the customized image recognition model provided by Baidu, and in the continuous operation of follow-up activities, the uploaded identified images were automatically added to the positive and negative sample database by classification through the AI identification result analysis module, It is constantly added to Baidu's customized model to continuously improve the accuracy of analysis.
Image recognition: retail end users upload display videos through e-store app, slice video files into image files through platform video processing, call Baidu customized image recognition interface, and identify whether the target image meets the requirements.
User use:
