Machine learning technique for object detection based on SURF feature
Amin Mohamed Ahsan, Dzulkifli Mohamad · International Journal of Computational Vision and Robotics · 2017
Local features that based on interest points have received a great interest in computer vision field and they play an important role in many applications, such as object recognition, tracking, and image retrieval. These features have proven to be invariant against the geometric and photometric transformations and proven to be robust under different types of image disturbances. Matching technique is usually employed in this field to recognise the object. Yet, it is not suitable for some applications such as searching for an isolated object, part-based object recognition, and object categorisation. A model for object detection with an artificial neural network (ANN) to overcome such shortages is proposed. Two datasets are prepared to be used for learning; one for human faces and the other for the cars. Features are extracted using speeded-up robust feature (SURF). The proposed model is evaluated using two benchmark datasets, Caltech101 and VOC2009. The obtained results are encouraging.