Local Image Feature Extraction using Stacked-Autoencoder in the Bag-of-Visual Word modelling of Images

Adaptive bag‐of‐visual word modelling using stacked‐autoencoder and particle swarm optimisation for the unsupervised categorisation of images

“…The results shown in [8] confirms the applicability of image feature extraction via Stacked-Autoencoder to the BOVW modelling process. Although unlike SIFT, Stacked-Autoencoder (and other Deep Feature Learning algorithms) do not provide scale and rotation Invariance representations [60], the results in [8] confirms that this deficiency is largely compensated for by the histogram representation approach of BOVW and the spatial pyramid included in the image modelling for the elimination of spatial incoherency.…”

“…In [8], Olaode and Naghdy demonstrated that stacked‐autoencoder image feature extraction's approach of reducing the number of features by taking advantage of the spatial redundancy during the spatial tiling resulted in considerable reduction in the categorisation time when compared with SIFT. Although the time taken is higher than the time taken to complete the unsupervised categorisation with SURF features due to the time taken to train the stacked autoencoder, the higher accuracy recorded by stacked autoencoder confirms its better efficiency.…”

“…In [8] the authors demonstrated that the opportunity to change the number of layers and the number of neurons in each layer of a Deep Learning algorithm allows the feature extraction process to be adaptable to the content diversity of the image collection during BOVW modelling, thus generating image feature vectors whose dimension guarantees optimum discrimination, unlike the fixed 128 dimensions of Scale Invariant Feature Transform (SIFT) and 64 dimensions of Speeded-Up Robust Feature (SURF) [57,58,59].…”

“…The semantic based annotation of images has been recognised as a viable means of bridging the semantic gap associated with Content Based Image Retrieval (CBIR) [1,2,3,4,5,6]. While the efficient annotation of a large image collection via supervised machine learning remains a challenge in computer vision and image retrieval [7,8,9], the application of Unsupervised Machine Learning principles such as K-means clustering, Self-Organising Maps or Hierarchical clustering [10,11] enables the image models computed from a given a large image collection to be grouped based on similarity [12,13,14,15], without the need for labelled training samples, therefore is a natural fit for achieving Image annotation [16,17,18]. However, to achieve such unsupervised categorisation, there is a need for an efficient and effective local image pattern representation and global image representations [19].…”

Stacked Autoencoder based Feature Compression for Optimal Classification of Parkinson Disease from Vocal Feature Vectors using Immune Algorithms

“…The following table represents the confusion matrix for a binary classifier and the next table represents the outcome confusion matrix of the proposed work. Accuracy decides the overall performance of the system by classifying the PD affected individuals from the healthy ones and the accuracy was determined in percentage, higher the percentage, higher the accuracy [28][29][30]. The classification accuracy for the datasets of this study was calculated using the below equation.…”