For Project Enquiry +91 9791 044 044

Learning Consistent Feature Representation for

Cross-Modal Multimedia Retrieval

ABSTRACT

In the proposed formulation, the basis matrices of different modalities are jointly learned based on the training samples. Moreover, a local group-based priori is proposed in the formulation to make a better use of popular block based features (e.g., HOG and GIST). Extensive experiments are conducted on four public databases: Pascal VOC2007, LabelMe, Wikipedia, and NUS-WIDE. We also evaluated the proposed algorithm with unpaired data. By comparing with existing state-of-the-art algorithms, the results show that the proposed algorithm is more robust.

EXISTING SYSTEM	PROPOSED SYSTEM
EXISTING SYSTEM:- The existing methods try to learn a common latent subspace, where the heterogeneity between two modalities is minimized so that cross-matching is possible. However, most of these methods require fully paired samples and suffer difficulties when dealing with unpaired data.	PROPOSED SYSTEM:- We also evaluated the proposed algorithm with unpaired data. By comparing with existing state-of-the-art algorithms the results show that the proposed algorithm is more robust and achieves the best performance. If the similarities between the images and documents can be computed directly, the computed similarity scores can be used to rank the retrieved results.
EXISTING ALGORITHM :- Canonical Correlation Analysis (CCA) Algorithm.	PROPOSED ALGORITHM :- Multiview Discriminant Analysis Algorithm (MvDA).
ALGORITHM   DEFINITION:- It aims at learning a latent space by maximizing the correlating relationships between two modality features. It has been widely used for multimodal problems.	ALGORITHM DEFINITION:- The most important reason is that the dimensionality of MvDA’s target feature space is the sum of the different modality features’ dimensions. The proposed algorithm just needs about 10 ms for processing 1000 test samples, which is the fastest algorithm compared to the others.
DRAWBACKS:- To solve optimally due to its high computational complexity, we propose a distributed and heuristic algorithm called D-vMDG. No top retrieval images retrieved.	ADVANTAGES:- An energy-efficient routing protocol based on dynamic programming is proposed for the constructed topology. Top images will be retrieved by giving query.