This disclosure describes techniques that, with user permission, utilize user actions as implicit feedback for image recognition. With user permission and express consent, post-recognition user actions are utilized to adjust image recognition model(s). An image recognition request and one or more image(s) are received. Image recognition techniques are applied to the received image(s) and results are provided to the user. Based on actions performed by the user post-recognition, various factors are inferred and utilized to train image recognition model(s) to improve the performance of the model for future searches. With permission, user affinity model(s) are trained that can provide customization of future search results for the particular user.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Zimmerman, Jared M. and Tran, Mai, "Improving Object Recognition Using Post-Recognition Actions", Technical Disclosure Commons, (August 28, 2020)