Mastering Computer Vision with TensorFlow 2.x

Classification of the image

After feature extraction through AlexNet, the classification of the image involves passing the feature vector through class-specific linear SVMs to classify the presence of the object within the region proposal. Using SVMs is a supervised machine learning method that assigns weight and bias to each of the feature vectors and then draws a line to separate the objects into specific classes. The separation is done by determining the distance of each vector from the line and then positioning the line so that the separation distance is maximal.