# AI use cases for Brandpose (escudero/brandpose)

Repo: [escudero/brandpose](https://github.com/escudero/brandpose) | Use cases: 3 | Page: https://dedupedai.com/use-cases/escudero-brandpose/

BrandPose is a personal project that estimates where shoppers look in a store, using an ordinary fixed security camera and open source AI models (YOLO11, Gaze-LLE, Depth Anything V2).

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Measure shelf attention from a fixed store camera video: https://dedupedai.com/use-cases/escudero-brandpose/measure-shelf-attention-from-fixed-camera-video/
- Render the full promo video with zones, depth and soundtrack: https://dedupedai.com/use-cases/escudero-brandpose/render-promo-video-with-zones-depth-and-soundtrack/
- Rebuild the store in Blender and render a shopper's view: https://dedupedai.com/use-cases/escudero-brandpose/rebuild-store-in-blender-and-shopper-pov/

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

Repo: [escudero/brandpose](https://github.com/escudero/brandpose) | Difficulty: Intermediate | Time: About 2 hours | Basis: From the README

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

**Who it is for:** A 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 problem:** Manually watching hours of store footage to see where shoppers look is slow and inconsistent.

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

### 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 to paste into your agent

```text
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.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/escudero-brandpose/measure-shelf-attention-from-fixed-camera-video/

## How to render the full promo video with zones, depth and soundtrack with Brandpose

Repo: [escudero/brandpose](https://github.com/escudero/brandpose) | Difficulty: Advanced | Time: About 3 hours | Basis: From the README

Rebuild the demo pipeline: extraction, zones, depth, render and synthesized audio.

**Who it is for:** A developer or content creator who wants to reproduce the demo video from the same public footage to study or share the method.

**The problem:** The demo video combines several stages (pose, gaze, depth, rules, render and audio) that are hard to chain without the documented order.

**The outcome:** A rendered video with gaze heat, zone ranking, 3D view and synthesized soundtrack, built from the promo scripts in scripts/promo/.

### Steps

1. Complete the setup in the Reproduce section: create the venv with `python3 -m venv .venv && .venv/bin/pip install -r requirements.txt` and download the source video.
2. Open docs/experiments/2026-10-14-promo-v2.md (in Portuguese) and follow its step-by-step for extraction, zones, depth, render and soundtrack.
3. Run the stages in order from scripts/promo/: extract_v2.py, depth.py, render_promo.py and audio.py, using the arguments shown in that document. (inferred, check the README)
4. Check the zones against config/zones_acti.json, which the README says were drawn for this camera.

### Prompt to paste into your agent

```text
Using the repo escudero/brandpose, follow docs/experiments/2026-10-14-promo-v2.md to run the promo pipeline in scripts/promo/ on the ACTi demo video. Report the output video path and the stage that failed, if any. Verify that the zone ranking in the rendered report matches the zones in config/zones_acti.json.
```

### Caveat

The README notes the process runs at about 3 frames per second on an M2 Pro, so a full video takes a long time, and the source footage is © ACTi Corporation.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/escudero-brandpose/render-promo-video-with-zones-depth-and-soundtrack/

## How to rebuild the store in Blender and render a shopper's view with Brandpose

Repo: [escudero/brandpose](https://github.com/escudero/brandpose) | Difficulty: Advanced | Time: About 4 hours | Basis: From the README

Calibrate the camera to marked corners, model the fixtures in Blender and render one shopper's view.

**Who it is for:** A 3D artist or computer vision researcher who wants to visualize tracked shopper paths inside a reconstructed store.

**The problem:** Tracked shopper paths and gaze rays are hard to interpret as numbers alone, and the camera-to-store geometry is not obvious from the image.

**The outcome:** A calibrated camera and fixture layout in config/, a Blender store model, and a rendered first-person view of one shopper with the side-by-side and shopper's-view videos.

### Steps

1. Complete the base setup with `python3 -m venv .venv && .venv/bin/pip install -r requirements.txt` and make sure the extraction from the promo run exists.
2. Install Blender 5.2 from https://www.blender.org.
3. Open docs/experiments/2026-10-18-blender-pov.md at the section #rodar (in Portuguese) and run the commands listed there for calibration, modeling and rendering.
4. Use scripts/blender/overlay.py to check that the fitted fixtures line up with the image before rendering the shopper's view with scripts/blender/pov.py.

### Prompt to paste into your agent

```text
Using the repo escudero/brandpose, follow docs/experiments/2026-10-18-blender-pov.md to calibrate the camera with scripts/blender/calib.py, then render the shopper's view with scripts/blender/pov.py for <person ID>. Report the output video path and the calibration result. Verify with scripts/blender/overlay.py that the fixtures match the image before accepting the render.
```

### Caveat

The README says the shopper's view is a reconstruction, with his position off by about 20 to 30 cm, and the ceiling, storefront and products are made up.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/escudero-brandpose/rebuild-store-in-blender-and-shopper-pov/