hand-crafeted features

hand-crafted featuresare manually engineered by data scientists.
Sometimes, we may see “learned features”, so what are learned features, and what are the differences between them.
Learned are features that are extracted from images by various algorithms.
give some examples to clarify the differences
If you are doing a task like clarifying the cats from dogs, you want to build a classifier, but you face a dilemma of how to input data to the classifier. There are two methods:
one is to input the raw pixel data directly. This issue requires vast data space; the other is to extract features from images to reduce feature space.
If you choose the second one, you have two more options
manually define a set of features and extract them, like edge detection, corners detection, etc. The problem with this approach is that nothing guarantees that the number of coroners is a good descriptor for classifying dog and cat images.
The alternative is to train a DL model to identify and extract features for the specified classification task. This is precisely what a convolutional network does. It searches for what features are best to classify.
(handcrafed features 就是人为的去检测特定的特征)