Run Inference (v1 - Latest)

post

https://api.ai.vision.tiliter.com/api/v2/agents/decision-video-summary/v1/inference

Summarise per-frame decisions across a video, sent as a contact sheet image of the sampled frames, into one overall answer, a short summary and key moments Submit an image to analyze using the Decision Video Summary agent.

Recent Requests

Time	Status	User Agent
Retrieving recent requests…

Loading…

Body Params

input_file_b64

string

required

length ≤ 1468006

Base 64 encoded input file data. Must be one of the following types: ['Image'].

question

string

required

length ≥ 1

The question to answer about the image

options

array of objects

required

length between 2 and 255

The fixed set of answers to choose between

Options*

frames

array of objects

required

length between 1 and 60

Per-frame results, one per tile of the contact sheet sent as input_file_b64, in the same order. input_file_b64 must be a contact sheet: a single image tiling the sampled video frames in time order, left to right then top to bottom, each tile labelled with its timestamp as m:ss so it matches time_seconds here. Get quick_choice, the quick probabilities and checks by calling decision-check-v1 on each frame with the same question and options; set thorough_choice and thorough_explanation from decision-review-v1 where a frame was reviewed.

Frames*

Headers

X-API-Key

string

required

Your API key for authentication

Responses

Language

Loading…

Response

Click Try It! to start a request and see the response here! Or choose an example:

application/json

200Successful Response

422Validation Error