Image-Similarity-Search-Milvus

Image Similarity search build on Milvus. Get similar images from the dataset. Can be used for augmentation, diffusion models and finding similar patterns in images.

1. Getting Started

Clone the repo:

git clone https://github.com/bigmb/mb_milvis.git

Pip install

pip install mb_milvus

2. Requirements

python>=3.6
numpy
pymilvus==2.0.0rc7
tensorflow
pandas
glob
argparse
cv2
pathlib
PIL
mb_utils
mb_pandas

Install all dependent libraries:

pip install -r requirements.txt

3. Run the file

Make sure you have the milvus 2.0.0rc7 docker-compose file and then Milvus docker running. (Refer to: https://milvus.io/docs/v2.0.0/install_standalone-docker.md)

wget https://github.com/milvus-io/milvus/releases/download/v2.0.0-rc7/milvus-standalone-docker-compose.yml -O docker-compose.yml

Start the docker container using

docker-compose up -d

Image embedings extraction methods available: [Resnet, Xception,VGG16,VGG19,InceptionV3, MobileNet] Default : Resnet

Running the search file

./milvus_search.py -path_loc "path_to_img_folder" -num "number of simliar images" -batch_size "Extraction batch size if needed" -collection_name "name_of_the_collection" -save_csv "Saving location of the final CSV output"

4. Running video

milvus_movie.mp4