Image Color Recognition and Optimization Based on Deep Learning

In order to solve the problem of image color recognition, this paper proposes a method recognition and optimization based on deep learning designs postprocessing framework word bag model (bow). The uses CNN features calculates feature similarity. sets with high similarity are input into classifier trained by bow clustering as preliminary retrieval results. results categories largest number images. experimental show that accuracy is 90.4% same data set classification category, which 10% higher than algorithm features. Conclusion. matching degree between be retrieved has been greatly improved.