Image Recognition in Digital Merchandising Industry
4 likes1,737 views
This document discusses how computer vision can be used in digital merchandising to improve inventory management. Computer vision uses algorithms to process images and extract information. It has evolved from early pattern recognition techniques to modern deep neural networks. For digital merchandising, computer vision can be used to automatically identify products in store images and track inventory levels in real-time, leading to benefits like reduced out of stock items and better service.
Image Recognition in Digital Merchandising Industry
- 1. Image Recognition in Digital Merchandising Industry
- 3. Digital Merchandising Benefits • Improved Service Level Monitoring • Reduced Data Collection Time • Out of Stock Reduction • Increase in SKU Availability • Merchandiser Self-discipline
- 4. What is Computer Vision? “Computer vision is a field that includes methods for acquiring, processing, analyzing, and understanding images in order to produce numerical or symbolic information” Wikipedia
- 5. Computer Vision Evolution Computer Vision Artificial Intelligence Image Processing Pattern Recognition Robotics Machine Vision 1960s 1970s Edge detection Line Labeling 1980s 1990s Face tracking Face recognition 2000s Feature-based recognition Image-based modeling Image-based rendering Region-based recognition High dynamic range Emotion recognition Robotic Vision Neural network based recognition 2010s Physics First digital camera Megapixel sensor CMOS chip SURF SIFT ORB BRIEF FREAK HOG GLOH Haar-cascades Viola-Jones FAST Latent SVM Deep NN LSTM NN MD LSTM NN RPROP FR NN BoW
- 12. Filtering
- 14. Demo
- 16. Computer Vision Algorithms • SURF • SIFT • FAST • FREAK • ORB • RANSAC
- 17. Computer Vision Libraries • OpenCV • EmguCV • FastCV • JavaCV
- 18. Results
- 19. 0 10 20 30 40 50 60 70 80 90 0.3 0.6 0.7 0.8 0.9 1 1.2 2 3 5 8 Recognitionquality Image size Recognition quality depending on image size
- 24. ?