Method for recognizing object in image

An apparatus for recognizing an object in an image, the apparatus comprising:

• a processor configured to
• receive an image including the object;
• output a preprocessed image by performing image enhancement processing of brightness operation processing and contrast operation processing and at least one image enhancement processing among denoise operation processing, deblur operation processing, high dynamic range operation processing, color tone mapping operation processing, defog operation processing, auto white balance operation processing, back light compensation operation processing, or decompression operation processing, on the received image, to increase a recognition rate of the object included in the received image;
• and
• recognize the object included in the image by inputting the preprocessed image to an input layer of an artificial neural network for object recognition.

A method for recognizing an object in an image, the method comprising:

• receiving an image including the object at a preprocessing module;
• outputting a preprocessed image by performing image enhancement processing on the received image, by the preprocessing module, to increase a recognition rate of the object included in the received image;
• recognizing the object included in the image by inputting the preprocessed image to an input layer of an artificial neural network for object recognition, by an object recognition module;
• inputting, to the object recognition module, a plurality of preprocessed reference images to the input layer of the artificial neural network for object recognition, each of the plurality of reference images including an object;
• and
• generating, by the object recognition module, feedback data associated with a recognition result of the object included in each of the plurality of preprocessed reference images,
• wherein the outputting of the preprocessed image by the preprocessing module includes outputting the preprocessed image by inputting the received image including the object to an input layer of an artificial neural network for preprocessing so as to increase a recognition rate of the object included in the received image, and
• wherein the outputting of the preprocessed image by the preprocessing module includes training the artificial neural network for preprocessing is trained, using the feedback data, to infer the plurality of preprocessed reference images that are optimized for an increase in object recognition rate through the artificial neural network for object recognition.

A method for recognizing an object in an image, the method comprising:

• receiving an image including the object;
• outputting a preprocessed image, by a preprocessing module, by performing image enhancement processing of brightness operation processing and contrast operation processing and at least one image enhancement processing among denoise operation processing, deblur operation processing, high dynamic range operation processing, color tone mapping operation processing, defog operation processing, auto white balance operation processing, back light compensation operation processing, and decompression operation processing, on the received image, so as to increase a recognition rate of the object included in the received image;
• and
• recognizing the object included in the image by inputting the preprocessed image to an input layer of an artificial neural network for object recognition, by an object recognition module.

An apparatus comprising:

• a first artificial neural network model, including a weight value, configured to input an image and output a preprocessed image by using the weight value;
• a second artificial neural network model configured to receive the preprocessed image and output a recognized object included in the preprocessed image;
• and
• a processor configured to process the first artificial neural network model and the second artificial neural network model,
• wherein the weight value is updated by a feedback data based on a recognition result of a reference object included in each of a plurality of preprocessed reference images through iterative training by inputting the plurality of preprocessed reference images, and
• wherein the weight value is determined to perform image enhancement processing in order to increase a recognition rate of the recognized object included in the preprocessed image input to the second artificial neural network model.