Start building right away with Twelve Labs API
Powerful and seamless video search infrastructure for your application. Quickly integrate SOTA and scalable video search in fewer than 20 lines of code.
APIs that will deliver your product, from start to end
Index & search flow
Simple 2-step integration to power your video solutions
Integrate SOTA video search infrastructure in fewer than 20 lines of code. Step 1 is to upload and vectorize your videos. Step 2 is to perform the task of your choosing.
Engines API
Select an AI engine
Twelve Labs default engine
Customer finetuned model
arrow
1
Index API
Upload and vectorize videos
Input: Raw video data
index-api
index-api
arrow
2
{{ Task }} API
a. Search API
Search through the vector space and retrieve target videos
search-api
b. Vector API
Return video data described as vectors in the vector space
vector-api
1
Dataset API
Provide a set of your own videos and custom labels
Input: Raw videos + Start-end time + Moment description
dataset-api
arrow
2
Finetune API
Train the Twelve Labs’ default engine to create your own engine
finetune-api
arrow
3
Deploy API
Add the finetuned model to the list of engines that you can select to index videos
deploy-api
Fine-tune & deploy flow
Quickly train your own model
Customize the Twelve Labs default engine to serve your own specific needs. Simply train our model with provided videos and labels, instead of creating a model from scratch.
Sneak peek of what you can do
World’s most cutting edge video understanding AI technology in a simple API call
  • img
    Search
    Make any video database searchable to instantly find key moments

  • img
    Vector
    Transform your video into a vector embedding that captures the semantics of the video

  • img
    Finetune
    Create a customized model for your domain-specific needs
Search the scene where Amy and her team are discussing marketing strategies in a conference room.
Request
import request

SEARCH_URL = "https://www.api.twelvelabs.io/v1/search"
QUERY = "Amy and her team are discussing marketing strategies in a conference room"
 
search_request = requests.post(SEARCH_URL,
                              data={"query": QUERY,
                                    "index_id": "61fb34ea795aae003158c700",
                                    "search_options": ["visual", "conversation", "people", "text_in_video"]},
                              headers={"x-api-key": "tlk_3E7JN5S1JJ4KXN2NVD0QA0NAV3AE"})

print(search_request.text)

    
API docs
arrowarrow
Interested in
making your videos searchable?
Next generation video understanding technology at your finger tips