https://github.com/kiarashvosough1999/imagefaceclustering
https://github.com/kiarashvosough1999/imagefaceclustering
Last synced: 7 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/kiarashvosough1999/imagefaceclustering
- Owner: kiarashvosough1999
- Created: 2022-03-31T06:45:52.000Z (over 3 years ago)
- Default Branch: master
- Last Pushed: 2022-04-17T21:40:54.000Z (over 3 years ago)
- Last Synced: 2025-01-29T13:45:47.733Z (8 months ago)
- Language: Jupyter Notebook
- Size: 1.28 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Image Face Clustering
cluster face images with and without **CNN**. I did some result analysis on the nootbook with visual diagrams and logs.
## Without CNN
On this approach I did not use any vision model architectures, I just extracted **Pixels** as features and transformed it from 3-dimensional to 2-dimensional.
## With CNN
On this approach I used any vision model architectures **VGG15**. It is considered to be one of the excellent vision model architecture till date.
Most unique thing about VGG16 is that instead of having a large number of hyper-parameter they focused on having convolution layers of 3x3 filter with a stride 1 and always used same padding and maxpool layer of 2x2 filter of stride 2. It follows this arrangement of convolution and max pool layers consistently throughout the whole architecture.
In the end it has 2 FC(fully connected layers) followed by a softmax for output. The 16 in VGG16 refers to it has 16 layers that have weights. This network is a pretty large network and it has about 138 million (approx) parameters.
> Pass `transformed_pca` to Cluster classes in order to use reduced features with PCA algorithm.
> Pass VGG16 model to `DataLoader.generate_raw_dataset` method to use CNN as features.
## Supported Cluster Algorithm
- [x] Kmeans
- [x] DBScan
- [x] Agglomerative (ward, complete, single, average)
- [x] Documented
- [x] Visual Analysis
## Kmeans
Kmeans and MiniBatchKMeans used and the results were compared.
## DBScan
DBScan were used with epsilon estimator which use **NearestNeighbour** to evaluate best epsilon. Also the effect of min samples on the amount of noisy detected datas were measured.
## Agglomerative
Four types ward, complete, single and average were used. Also the best threshold was measured and the differences of 4 types were analysed.