|
--- |
|
license: mit |
|
datasets: |
|
- rmit-denominator/BloomSage-Feature_extractor |
|
language: |
|
- aa |
|
metrics: |
|
- accuracy |
|
library_name: keras |
|
pipeline_tag: image-classification |
|
tags: |
|
- biology |
|
--- |
|
|
|
## BloomSage Flower Classification and Recommendation Models |
|
- The repository contains 3 flower classification model and 1 feature extractor model for flower recommendation. |
|
- For more specific instruction, please visit https://github.com/rmit-denominator/bloomsage-ml |
|
|
|
### Applications |
|
1. Object recognition |
|
2. object classification |
|
3. Building a recommender system |
|
### Selected models |
|
- For classification, we use the basic structure of Artificial Neutral Network (ANN) and Convolutional Neutral Network (CNN). |
|
- For feature extractor, we constructed a Convolutional Neural Network (CNN) to extract feature vectors from user preferences image |
|
- Apply a K-Means unsupervised machine learning model to cluster the reference image's feature vector with those of the images in our database. |
|
### Limitations |
|
- Since our target customer are small flower shops, we just use a sample of 8 flower species with 16362 images. |
