I would like to interpret a trained model for an image recognition system and understand which part of the feature vector(the last layer that I am using for cosine similarity or matching with enrollments) is for color information.
For example, if I have two images like this:
I want to know which part of the feature vector is different and this difference is related to color differences.
Dear Jalil Nourmohammadi Khiarak ,
Although Deep Convolutional Neural Networks have been widelyapplied for object recognition, most of the works have often based their analysison the results generated by a specific network without considering how theinternal part of the network itself has generated those results. The visualizationof the activations and features of the neurons generated by the network can helpto determine the best network architecture for our proposed idea. By the appli‐cation of deconvolutional networks and deep visualization, in this work, wepropose an analysis to determine which kind of images with different color spec‐trum provide better information to generate a better accuracy of our CNN model.The focus of this study is mostly based on the identification of diseases andplagues on plants. Experimental results on images with different diseases fromour Tomato disease dataset show that each disease contains valuable informationin the infected part of the leaf that responds differently to other uninfected partsof the plant (PDF) Spectral Analysis of CNN for Tomato Disease Identification. Available from: https://www.researchgate.net/publication/318134312_Spectral_Analysis_of_CNN_for_Tomato_Disease_Identification [accessed Jan 03 2024].
Conference Paper Spectral Analysis of CNN for Tomato Disease Identification
Regards,
Shafagat
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python