Dealing with all unstructured data, such as reverse image search, audio search, molecular search, video analysis, question and answer systems, NLP, etc.

Report Bug or Request Feature

Reverse Image search	Chatbots	Chemical structure search

About Milvus Bootcamp

Embed everything, thanks to AI, we can use neural networks to extract feature vectors from unstructured data, such as image, audio and vide etc. Then analyse the unstructured data by calculating the feature vectors, for example calculating the Euclidean or Cosine distance of the vectors to get the similarity.

Milvus Bootcamp is designed to expose users to both the simplicity and depth of the Milvus vector database. Discover how to run benchmark tests as well as build similarity search applications like chatbots, recommender systems, reverse image search, molecular search, video search, audio search, and more.

Solutions

Run locally

Here are several solutions for a wide range of scenarios. Each solution contains a Jupyter Notebook and a Docker deployable solution, meaning anyone can run it on their local machine. In addition to this there are also some related technical articles and live streams.

Solutions	Have fun with it	Article	Video
Reverse Image Search Build a reverse image search system using Milvus paired with YOLOv3 for object detection and ResNet-50 for feature extraction.	- Jupyter notebook - Quick deploy - Object detection	- 中文 - English	- 中文
Question Answering System Build an intelligent chatbot using Milvus and the BERT model for natural language processing (NLP).	- Jupyter notebook - Quick deploy	- 中文 - English	- 中文
Recommender System Build an AI-powered movie recommender system using Milvus paired with PaddlePaddle’s deep learning framework.	- Jupyter notebook - Quick deploy	- 中文
Molecular Similarity Search Build a molecular similarity search system using Milvus paired with RDKit for cheminformatics.	- Jupyter notebook - Quick deploy	- 中文	- 中文
Video Similarity Search Build a video similarity search engine using Milvus and a VGG neural network. Also paired with YOLOV2 & ResNet-50 to detect object in video.	- Jupyter notebook - Quick deploy - Object detection	- 中文 - English
Audio Similarity Search Build an audio search engine using Milvus paired with PANNs for audio pattern recognition.	- Jupyter notebook - Quick deploy	- 中文
Text Search Engine Build a text search engine using Milvus and BERT model.	- Jupyter notebook - Quick deploy	- 中文	- 中文
DNA Sequence Classification Build a DNA sequence classification system using Milvus with k-mers & CountVectorizer.	- Jupyter notebook - Quick deploy	- 中文

Live Demo

We have built online demos for reverse image search, chatbot and molecular search that everyone can have fun with.

Benchmark Tests

The Benchmark Test contains 1 million and 100 million vector tests that indicate how your system will react to differently sized datasets.

1 million benchmark testing

We extracted one million vectors from the SIFT1B Dataset for accuracy tests and performance tests. Through this test, you can learn the basic operations of Milvus, including creating collections, inserting data, building indexes, searching, etc.

100 million benchmark testing

We extracted 100 million vectors from the SIFT1B Dataset for accuracy tests and performance tests. Through this test, you can learn the basic operations of Milvus, including creating collections, inserting data, building indexes, searching, etc.

Collaborations

Milvus_and_ONNX

Build a reverse image search system with Milvus using various AI models in collaboration with the Open Neural Network Exchange (ONNX).

Contributing

Contributions to Milvus Bootcamp are welcome from everyone. See Guidelines for Contributing for details.

Supports

Join the Milvus community on Slack to give feedback, ask for advice, and direct questions to our engineering team. We also have a WeChat group.