Brand detection
Brand and logo detection for video
Paste a YouTube, TikTok or Instagram link and see every brand in the video: when its logo shows up, how long it stays on screen, and each time someone says its name.
Brands in the picture, the speech and the caption
A sponsor is usually on the product, said in the voiceover and tagged in the caption. NetraFlow finds all three and keeps them apart.
- On screen
- Every logo and wordmark it can read, with how long it showed and a box on the frame where it was found.
- Said aloud
- Every time the brand is named in speech, with the time and the words around it.
- In the post
- Brands named in the title, the caption or the hashtags.
- What it sits on
- Name surfaces such as sneakers, a jersey or a can, and only logos on those count. Useful when you paid for the shoe and the sticker on the wall behind it should not count.
How long each logo is on screen
A real Instagram reel checked for Skechers on sneakers. The bar shows when the logo was on the shoe, the diamond when it was said, and each tile is a frame.
Brand detection
1 brand7s total exposureSurface: sneakers
0:000:15
audio mention
How it works
- 01
Paste a link
A YouTube video or Short, a TikTok, an Instagram reel, or a file you upload. Name brands or surfaces to narrow it down, or leave them out to find every brand.
- 02
NetraFlow checks the picture, sound and caption
It finds, names and times every logo on screen, and listens and reads for brand names at the same time.
- 03
Get the evidence
Screen time per brand, the moments it appears, and the frames that prove it.
For developers: the same result from the API
Send a link, get the same brands, times and boxes back for your own reports or alerts. The sponsor in this reel, Greptile, was on screen for 7.8 seconds, said once, and tagged in the caption.
Request: POST /v1/jobs
curl -X POST https://api.netraflow.com/v1/jobs \
-H "X-Api-Key: $NETRAFLOW_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"url": "https://www.instagram.com/reel/DeTDjNTvEXe/",
"capabilities": ["brands"]
}'Result: GET /v1/jobs/{job_id}, once it finishes
{
"data": {
"job_id": "job_Hv2zJeryHCVn-wKt",
"status": "completed",
"platform": "instagram",
"video_title": "Video by tech_wizzdom",
"video_duration_seconds": 37,
"credits_charged": 4,
"results": {
"transcription": { … },
"brands": [
{
"name": "greptile",
"raw_name": "Greptile",
"confidence": 0.9,
"total_appearances": 3,
"total_visible_duration": 7.808,
"mark_seconds": 7.808,
"peak_marks": 1,
"first_seen": 23.757,
"last_seen": 31.565,
"appearances": [
{
"source": "visual",
"verified": true,
"timestamp_start": 23.757,
"timestamp_end": 31.565,
"visible_duration": 7.808,
"confidence": 0.9,
"bounding_box": {
"x": 0.14722222222222223,
"y": 0.4421875,
"width": 0.7046296296296296,
"height": 0.09791666666666667
},
"frame_detections": […]
},
{
"source": "audio",
"context": "Mentioned in speech",
"timestamp_start": 24.125,
"timestamp_end": 24.765,
"confidence": 0.8074341,
"speech_context": "PRs. I use Greptile, the code reviewer"
},
{
"source": "metadata",
"context": "Mentioned in video metadata",
"confidence": 0.5
}
]
},
{
"name": "louis vuitton",
"raw_name": "Louis Vuitton",
"confidence": 0.95,
"total_appearances": 2,
"total_visible_duration": 2.636,
"mark_seconds": 2.636,
"peak_marks": 1,
"first_seen": 7.374,
"last_seen": 35.569,
"appearances": […]
},
{
"name": "magsafe",
"raw_name": "Magsafe",
"confidence": 0.8,
"total_appearances": 2,
"total_visible_duration": 3.403,
"mark_seconds": 3.403,
"peak_marks": 1,
"first_seen": 5.639,
"last_seen": 14.715,
"appearances": […]
},
{
"name": "monster",
"raw_name": "Monster",
"confidence": 0.79,
"total_appearances": 1,
"total_visible_duration": 3.203,
"mark_seconds": 5.472,
"peak_marks": 2,
"first_seen": 0,
"last_seen": 3.203,
"appearances": […]
},
{
"name": "ralph lauren",
"raw_name": "Ralph Lauren",
"confidence": 0.9,
"total_appearances": 2,
"total_visible_duration": 3.537,
"mark_seconds": 3.537,
"peak_marks": 1,
"first_seen": 11.044,
"last_seen": 18.018,
"appearances": […]
}
],
"metadata": { … },
"storyboard": { … },
"confidence_summary": {
"high": 4,
"medium": 1,
"low": 0,
"total": 5
}
}
}
}… marks what was left out. Questions about brand detection
Do I have to tell it which brands to look for?
No. Leave it empty and it reports every brand it finds, on every plan. Name brands to keep only those, or name surfaces such as shirts and sneakers to look only there.
How is screen time measured?
It is the seconds a brand was on screen at all, so two logos showing at once count once. A second number counts each logo on its own, to show how heavily the screen was covered.
Can it tell the sponsor from a logo that happens to be in shot?
It reports both, with the evidence for each. A sponsor is usually seen, said and tagged; a logo on a passing car is only seen, briefly. You can also ask "Which brand is this video promoting?"
How accurate is it?
Every brand comes with a confidence score and the frames where it was found, so you can check it yourself. A tiny, blurred or covered logo can be missed, so look at the frames before a result settles a contract.
Which videos does it work on?
Public YouTube videos and Shorts, TikToks, Instagram reels and video posts, and files you upload. To check every new video from a YouTube channel or a TikTok creator, set up a follow.
Find the brands in your next video
Paste a link, or ask Claude. New accounts get 50 free credits. No card needed.