DEDUPED Hot AI repos, turned into use cases Get 5 a day

How to measure shelf attention from a fixed store camera video with Brandpose

Run the gaze pipeline on a fixed-camera clip to rank which shelf zones draw attention.

Authorization notice. Brandpose is dual-use. Only use it on software, sites and systems you own or have explicit permission to test.

Who it is forA retail analyst or store operator who has footage from a fixed security camera and wants a rough view of which shelves shoppers look at.

The problemManually watching hours of store footage to see where shoppers look is slow and inconsistent.

The outcomeAn estimated gaze count, shopper count and attention share per zone for the first 60 seconds of the clip, written to outputs/v2.

What you need first

  • Python 3.13
  • FFmpeg installed and on the system
  • About 10 GB of free disk space
  • A machine able to run the models; the README reports tests on a Mac M2 Pro with 16 GB using MPS acceleration
  • Blender 5.2 only if you want the 3D store and shopper's view

Steps

  1. Clone the repo, then run python3 -m venv .venv && .venv/bin/pip install -r requirements.txt inside it.
  2. Download the demo video with .venv/bin/yt-dlp -f "bv*[height<=1080][ext=mp4]+ba[ext=m4a]" -o "data/videos/youtube/yt_%(id)s.%(ext)s" "https://www.youtube.com/watch?v=-1bRhYjw1qE", or place your own fixed-camera video in data/videos/. (inferred, check the README)
  3. Run cd scripts and then ../.venv/bin/python gaze_sim_v2.py ../data/videos/youtube/yt_-1bRhYjw1qE.mp4 ../outputs/v2 --crop --model gazelle_dinov2_vitl14_inout --smooth 0.5 --suppress-self --track-every-frame --max-seconds 60. (inferred, check the README)
  4. Open the results in outputs/v2 and compare the zone ranking with your own rough count of a few minutes of footage.

Prompt for your agent

Using the repo escudero/brandpose, run scripts/gaze_sim_v2.py on the fixed-camera video at <video path> with --max-seconds 60, using the flags in the README. Report the number of estimated gazes, shoppers and zones with attention, and the output folder. Verify that outputs/v2 contains results and that the attention figures are labeled as model estimates, not ground truth.

Caveat. The README states there is no ground truth, so the percentages are model estimates and were not checked against manual annotations.

Share: X LinkedIn Hacker News Email Copy link

Disclaimer

Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked "Seen in the wild", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email [email protected] and we will look at it promptly. Full disclaimer page