| Autor: |
Varun, Akash, Rohit, Trishwanth, Maheshwari, K. M. Uma, Chaitanya, A. N. V., Prateek, T. S. S. |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2024, Vol. 3075 Issue 1, p1-6, 6p |
| Abstrakt: |
We describe an accurate and scalable method for face clustering. A group of faces is classified based on their prospective identities. The task is presented as a link prediction problem, where two faces are linked if they have the same identity. The fundamental premise is that the local context in the feature space surrounding an instance (facial) contains crucial knowledge about the interconnectedness of this occurrence with its neighbors. The key contribution of this work is at the beginning the face is detected using the computer vision-based detection algorithm and extracted the features using VGG16 the pre-trained model. By applying the Euclidian distance algorithm, the similarity score of the face is obtained. Finally, similar faces are clustered using unsupervised learning algorithms such as K-means and Density-Based Spatial Clustering of Applications with Noise(DBSCAN). Furthermore, these two clusters are compared and DBSCAN gives a better silhouette score for DBSCAN than the K-means score. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
