Research on Feature-based Image Recognition based on Convolutional Neural Network
Qi Ma, Zeyu Yang, Qingyao Zhang, Yejia Yang, Haibin Ran, Yuhao Bian · 2023
Recently, image identification has achieved promising performance and become the most essential component in real-life image-based applications through training a machine learning model. However, existing methods extract the features and the identification accuracy is almost depended on these features with extractions process, which is the fundamental process for learning models. Existing methods are almost concentrated on the utilization of these feature and ignore the process of extracting procedure. As far as we can concern, the image features contain numerous information and can assist the learning model to enhance the identification accuracy. In this work, we initially utilize a convolutional neural network to obtain the features in the feature space. Subsequently, an attention mechanism and the feature clustering are applying to optimize the extracted features. Finally, we evaluate our proposed model with most used methods in the identical environment setups. From our extensive experimental results, we can observe that our model outperforms compared methods with reasonable computation costs.