Multiple Instance Learning for Car Brand Classification in the Leboncoin Online Marketplace
Master thesis
Permanent lenke
http://hdl.handle.net/11250/2615853Utgivelsesdato
2017Metadata
Vis full innførselSamlinger
Sammendrag
This research project applies machine learning to a large-scale dataset of car images fromthe marketplace website Leboncoin. The project develops a model that classifies imagesfrom sales ads of cars with the brand of the car. The model combines the predictions of allthe images within an ad to achieve an ad-wise classification.
The motivation behind automatic image classification is to increase accuracy, identifymislabeled items and guide the user through the registration process of a sales ad. Themethods developed in this project can be applied to a variety of visual classification tasks,such as other e-commerce sites, medical imagery and video classification.
The project uses transfer learning to obtain faster training and higher accuracy due to alimited size of the dataset. The combination of predictions draws inspiration from multipleinstance learning and shows that averaging of the predictions is a powerful bag-of-instanceclassifier.
The classifier developed in this project gives an accuracy of 0.823 and 0.945 for image-wiseand ad-wise respectively.