Online Face Recognition
Overview
Online Face Search detects faces in an image and searches the web for visually similar faces. You upload an image or open one from a Workspace, select a detected face, and run a search that returns matches with similarity scores. Outputs include a grid of best matches and a downloadable ZIP of matched images. Results require review because similarity is not identity confirmation.
When to Use This Application
- You have a face image and need to find visually similar images on public web sources.
- You need lead generation when you have limited identifiers beyond an image.
- You need to compare multiple faces in one image by selecting each detected face.
- You need a ZIP package of candidate matches for offline review or disclosure workflows.
- You need to store the input image in a workspace as part of an investigation record.
Before You Begin
- Prepare an image that contains at least one clear, visible face.
- If you plan to save the image to a workspace, confirm you have write access to the destination folder.
- Confirm your environment allows outbound access to the online face search service.
Step-by-Step Walkthrough
Step 1 — Provide an image
Provide an image using one of these methods:
- Upload a file using the file upload button.
- Open from workspace by right-clicking an image file and launching the app from the context menu. After the image loads, it appears in the main area.
Step 2 — Detect faces
Select Detect Faces in the left sidebar. The app scans the image and highlights detected faces:
- Each face is outlined with a red border.
- Each face receives a number (Face 1, Face 2, …).
- A message confirms how many faces were found. If no faces are detected, the app shows an error message. Load a different image and try again.
Step 3 — Select a face
If the image contains more than one detected face, use the dropdown selector to choose which face you want to search.
Step 4 — Configure search settings
Configure the search settings in the left sidebar before you run the search.
Step 5 — Save image to workspace (optional)
Save the input image to a workspace when you want to preserve it as evidence or share it with your team.
- Enable Save image to workspace.
- Select the destination Workspace from the dropdown.
- Enter the file path where the image should be stored, for example
Faces/my-image.png. The app saves workspace images inPNGformat.
Step 6 — Search the web for a face
Select Search Web for Face. The app:
- Analyzes the selected face
- Sends it to the online face search service
- Shows a loading spinner with a scanning message
- Displays results when the search completes The search can take up to ~100 seconds depending on service response time.
Step 7 — Review results
Results appear under Best Matches in a grid. Each match includes:
- A face thumbnail
- The source URL
- A similarity score
Use the per-result context menu (
⋮) to interact with an individual match.
Understanding the Output
Online Face Search outputs a ranked grid of web matches.
- Face detection overlay
- Red borders and face numbers show which faces the app detected in the input image.
- Best matches grid
- Each result includes a thumbnail, source URL, and a similarity score. Higher similarity indicates a closer visual match, but it does not confirm the person’s identity.
- Similarity score
- Use the score to prioritize review. Confirm candidates by opening the source images and checking contextual evidence.
Saving and Exporting Results
Online Face Search supports two output actions:
- Save the input image to a Workspace (optional)
- What it does: stores the input image in a workspace folder
- Where it ends up: specified folder path in the selected workspace
- Settings: workspace selection and file path
- Note: saved images are stored as
PNG
- Download matched images as ZIP
- What it does: downloads all matched face images as a single ZIP file
- Where it ends up: your local device downloads
- Settings: none Select Download Images to download the ZIP.
Tips for Best Results
- Use a clear, well-lit image with faces that are not heavily obscured.
- Run Detect Faces before adjusting search settings so you confirm face selection.
- Use
Facenetas a default model. TryArcFaceorFacenet512when you need a different trade-off in matching behavior. - Increase Max Results when you need broader coverage and the first run returns few matches.
- Save the input image to a workspace when you need auditability and team review.
- Retry after timeouts. The online search relies on an external service and can fail due to connectivity or service delays.
Known Limitations
- Web results depend on an external service. Timeouts and connectivity failures can occur.
- Similarity is not identity confirmation. False positives can occur, especially with low-quality images.
- If no faces are detected, you must provide a different image or improve image quality.
- Workspace-saved images are stored as
PNG, which may change file format from the original upload.