Skip to main content
POST
/
rtstream
/
{stream_id}
/
search
Search RTStream scene index
curl --request POST \
  --url https://api.videodb.io/rtstream/{stream_id}/search \
  --header 'Content-Type: application/json' \
  --header 'x-access-token: <api-key>' \
  --data '
{
  "query": "person walking",
  "scene_index_id": "scene-idx-12345",
  "result_threshold": 10,
  "score_threshold": 0.5,
  "dynamic_score_percentage": 20,
  "stitch": true,
  "filter": [
    {}
  ],
  "rerank": false,
  "rerank_params": {}
}
'
{
  "success": true,
  "data": {
    "results": [
      {
        "start": 1700000000,
        "end": 1700000010,
        "text": "matching scene text",
        "score": 0.95,
        "scene_index_id": "scene-idx-12345"
      }
    ]
  }
}

Documentation Index

Fetch the complete documentation index at: https://docs.videodb.io/llms.txt

Use this file to discover all available pages before exploring further.

Search for specific content within a real-time stream’s scene index using semantic search.
  • query and scene_index_id are required
  • result_threshold limits the number of results (default: 10)
  • score_threshold filters by minimum relevance score
  • Enable rerank for improved result ordering
  • Use stitch to merge adjacent matching segments (default: true)

Create Scene Index

Set up scene indexing first

Event Detection

Build event detection patterns

Authorizations

​
x-access-token
string
header
required

API key for authentication (sk-xxx format)

Path Parameters

​
stream_id
string
required
Example:

"rts-12345"

Body

application/json
​
query
string
required
Example:

"person walking"

​
scene_index_id
string
required
Example:

"scene-idx-12345"

​
result_threshold
integer
default:10
Example:

10

​
score_threshold
number
Example:

0.5

​
dynamic_score_percentage
integer
default:20
Example:

20

​
stitch
boolean
default:true
Example:

true

​
filter
object[]
​
rerank
boolean
default:false
Example:

false

​
rerank_params
object

Response

200 - application/json

Search results

​
success
boolean
Example:

true

​
data
object