Streamingo Face Blur

By Streamingo.ai

Overview

Automates face blurring in video and image files to streamline privacy compliance and data anonymization. Streamingo detects and obfuscates faces across media without destroying context, helping developers and data teams safely share, process, or publish visual media while adhering to GDPR, CCPA, and strict privacy standards.

Tools

blur_image

ChatGPT

Anonymize/blur a new image using the provided image URL in a folder that is part of a project. Args: input_url (str): image URL (publicly accessible). project_id (str): project id where image will be created. organization_id (str): organization id. parent_folder_id (str): folder id (must be an IMAGE folder) where image will be created. name (str): display name for the image (default: "image"). height (int): image height in pixels; use 0 if unknown (backend may infer). width (int): image width in pixels; use 0 if unknown (backend may infer). toggle_hairblur (bool): blur face only if False, else full head including hair. Returns: str: Success message with image_id. Instructions: 1. Ask if user wants to blur only the face or head including hair; set toggle_hairblur accordingly. 2. Use fetch_folders(project_id=..., type='IMAGE') to get image folders; user must pick project and folder before calling this.

blur_video

ChatGPT

Anonymize/blur a new video using provided input url in a folder that is part of a project. Args: input_url (str): video url, can be a youtube url or any other video url. project_id (int): project id where video will be created. parent_folder_id (str): folder id where video will be created. toggleHairblur (bool): Will only blur face if False, else full head remove_audio (bool): Output will contain no audio if True modifyAudio (bool): output will contain anonymized audio if True enableReview (bool): User will be able to edit and review output if True selectiveBlur (bool): User will be able to use Selective blur feature if True Returns: str: videoId thats created Instructions: Follow these steps one by one. 1. Ask if user wants to blur only the face or head including hair? Pick toggleHairblur argument based on user response. 2. Ask user if they want to remove the audio, or anonymize the audio or keep same audio in output, Select remove_audio and modifyAudio arguments based on user response. 3. Ask if user wants to review the output for further edits later? Pick enableReview argument based on user response. 4. Ask user if they want to use selective blur feature. It allows user to select/deselect individual faces and blur again later. 5. If we are able to submit video succesfully and he has picked enableReview or selectiveBlur feature tell this to user: - To review the video you would need to login to ananymize platform.

fetch_folders

ChatGPT

Returns a list of recent folders inside a project. Args: project_id (str): project id page (int): page number for api results limit (int): limit of results to fetch type: can be VIDEO or IMAGE if its a video folder or image respectively. Returns: List: List of folders Instructions: 1. use page=1 and limit=5 initially 2. Select the first project as the default and present user in following format. 3. - List all available projects from current page - Explicitly ask user `I have selected this project_name, should I proceed or you want to select different project. - If user ask for another project that's not mentioned in list, load second page and follow step 3 from start. 4. Never proceed without user confirmation

fetch_image_by_image_id

ChatGPT

Fetch a single image by image_id. Use this to get details or status of an image. Args: image_id (str): image id of an existing image. Returns: Image: image object (includes status, inferred_url when ready, etc.). Instructions: If user asks for download/output: use inferred_url when inference status is SUCCESS. If user asks for status: check inference status and download status; if FAILURE, log_file may contain details.

fetch_images

ChatGPT

Fetch images inside an image folder. Use this to list existing images in a folder. Args: parent_folder_id (str): folder id (IMAGE type) to list images from. Returns: List: list of image objects in the folder.

fetch_projects

ChatGPT

Returns a list of recent projects. Args: project_id (str): project id page (int): page number for api results limit (int): limit of results to fetch Returns: List: List of folders Instructions: 1. use page=1 and limit=5 initially 2. Select the first project as the default and present user in following format. 3. - List all 5 available projects from current page - Explicitly ask user `I have selected this project_name, should I proceed or you want to select different project. - If user ask for another project that's not mentioned in list, load second page and follow step 3 from start. 4. Never proceed without user confirmation

fetch_video_by_video_id

ChatGPT

Fetch single video by a videoId. Use this tool if we want to any detail related to a video. Args: video_id (str): video id of existing video. Returns: Video: returns video object Instructions: If user asks for download video output use following steps: - Proceed only if inference status is success else tell user about the status. - return inferedUrl as the output video url. If user asks for status of the video use following steps: - check download status, if not complete update user about its status else proceed further. - check inference status, if its SUCCESS, return inferredUrl as output, else update user about its status. - if any status is Failed, it means video filed to proceed, you can return logFile for logs.

fetch_videos

ChatGPT

Fetch videos inside a folder. Use this to get existing videos details. Args: parent_folder_id (str): folder id where video will be created. Returns: str: videoId of the created video

generate_embedd_link

ChatGPT

Generates embedd link of the blurred media if user wants to share and view it. This embedd url only opens inside anonymzation official platform and it won't download the media. Args: media_id (str): id of the media. expiry_days (int): expiry of the embedd link url in days media_type (str): type of the media (video, image) Returns: str: embedd_link of the Instructions: Ask the use about the validity of media in days and pick expiry_days argumant based on user response.