Object Detection and Classification: A Joint Selection and Fusion Strategy of Deep Convolutional Neural Network and SIFT Point Features

Published in Multimedia Tools and Applications (Springer), 2018

Authors:
Muhammad Rashid, Muhammad Attique Khan, Muhammad Sharif, Mudassar Raza, Muhammad Masood, Farhat Afza


This work presents a hybrid framework for object detection and classification by combining deep convolutional neural network (CNN) features with SIFT-based handcrafted features. The proposed approach performs feature selection and fusion to improve classification performance by leveraging both deep and traditional feature representations.


Contributions 📃

  1. A hybrid framework combining CNN features and SIFT features.
  2. Joint feature selection and fusion strategy for improved performance.
  3. Enhanced object detection and classification accuracy.
  4. Demonstration of the effectiveness of combining deep and handcrafted features.

📖 Citation (BibTeX)

@article{rashid2018object,
  title     = {Object Detection and Classification: A Joint Selection and Fusion Strategy of Deep Convolutional Neural Network and SIFT Point Features},
  author    = {Rashid, Muhammad and Khan, Muhammad Attique and Sharif, Muhammad and Raza, Mudassar and Masood, Muhammad and Afza, Farhat},
  journal   = {Multimedia Tools and Applications},
  year      = {2018},
  publisher = {Springer},
  doi       = {10.1007/s11042-018-7031-0}
}

Paper PDF Code