Random Forest Ensemble Learning for Object Recognition Using RGB Features Along Object Edge

In object recognition, the required features to represent an object are dictated by the nature of the problem at hand, and thus extraction of features from an object varies from task to task. State-of-the-art object detectors typically use the shape information as a low level feature representation to capture the local structure of an object. This paper presents the concept of merging the shape (in the form of edge) and color information, which could lead to significant performance improvement in object detection. In real life scenarios, we often need to recognize objects that are captured at different angles. In this paper, we propose a method that capture the RGB colors along the image’s edge as importance features for object recognition. We test our method on NEC animal object dataset. Our simple yet effective approach is able to offer good image classification results when we use it in conjunction with random forest ensemble learning method for classification. Keywords—Image classification, RGB features along edge, Ensemble learning, Random forest