CODEWITHSUNDEEP
algorithmmachine-learning
Ashwin
Ashwin

Reputation: 431

Training and Testing of Algorithm in OD

Hi,
I am working on image processioning Specific on object detection algorithms .
Frequently I am coming across two terms called as training and testing of image .
I read some papers also about this, but not fully understood these terms in object detection.

Can some of us explain in easy language about these terms :

1)Training of Algo
2)Testing of Algo

Tr,
Ashwin

Upvotes: 0

Views: 469

Answers (1)

Nico Schertler
Nico Schertler

Reputation: 32627

Most machine learning algorithms can be split into three phases. The training phase, the validation phase (optional) and the testing phase.

During training, the parameters of the algorithm are learned from a given training data set (in the case of supervised learning). Usually, this training data set consists of the algorithm's input and its output (so what the algorithm should do with a certain input). In object detection, this can be labeled images (what regions of an image correspond to what object). Since there are various algorithms, the learned parameters vary as well. In the case of neural networks, the edges' weights are being learned.

Testing is then the actual execution of the algorithm. You provide input and the algorithm calculates the output based on the learned parameters from the training phase. For object detection, the input might be an image, and the output might be each pixel's label.

Validation can be used after training in order to generalize parameters (i.e. avoid over-fitting).

Upvotes: 2

Related Questions