Logo Matcher

Home - Editor - Simple_Upload

Home Page

The logo matcher is a Project VIC prototype technology developed to assist investigators identify unknown logos. This logo matcher has a database of 10K different logos for matching. This includes many international brands, national brands within the United States, and regional brands from within the United States.

Occasionally investigators encounter an image where a human created mark (e.g., a logo, writing, design) may be the only clue to identifying additional information about the context of the image, like identifying identities and locations. Sometimes these images contain human created marks in the background of the image that are blurry, occluded, or both.

The logo matcher extracts these marks and attempts to match them against a database of curated logos. The tool may detect a logo and not identify the name. This is because Project VIC has not yet named that logo in the database.

 

 

 

 

The investigator can also supply partial logos for matching. The Logo Matcher requires around 50% of the original logo for matching.

 

 

 

 

Project VIC International continues to curate the logo matching database. In the image above, the logo was matched to the Adidas3 class with a confidence of 1.0. This means that the matcher has 100% confidence that it is a match. Strong matches are usually between 99-100%. However, investigators should review other matches as the matcher will sometimes provides matches at a much lower confidence.

The matcher has the ability to match logos collected from different angles, colors, lighting, and material.

At this time, the Logo Matcher only supports one image frame submission at a time. We can try to identify marks within a video file if you send it to us for offline processing. You can use a common snipping tool to capture a frame from your video and submit it to the logo matcher.

Directions for use:

Do not upload CSAM to this service. Please occlude all disturbing content before submission as Project VIC staff may see your submission.

The user may choose to pre-edit an image frame before submission to the Logo Matcher for processing. If this is the case, the user clicks the "Editor" link at the top of the page which takes them to the Toast UI Image Editor page.

 

 

 

 

Select "Load" and then upload the source image into the editor. Select one of the tools in the editor to make changes, such as crop, or blur, to make the necessary edits.

The user must select "apply" to finalize the edits before submission. All of the edits are completed in the user's browser.

When the edits are complete, the user should name the file and click the "upload to server" button. The user can also download the edited image by selecting the "download" button.

 

 

 

 

Alternatively, the user may also choose to submit the entire image frame for processing.

Select "upload_simple" at the top of the page to do this.

Choose a file, then hit the upload button.

The logo matcher takes 1-3 seconds to process each frame depending on how many logos are found in the image. After processing is complete, the logo matcher returns the finding to the user via a list of processed submissions. Find the recent submission or a particular named submission and select the filename.

 

 

 

 

The application will then present the submitted image, detected human created marks, and matched logos.

 

 

 

 

The logo matcher will not find everything, and it will not match everything. However, we hope that this service will help investigators find contextual clues about their image.