Thread by @skalskip92

time analysis with computer vision

- blurring faces
- detection and tracking
- smoothing detections
- filtering detections by zone
- calculating time

let me know if you want me to explain anything else. ;)

code:

↓ read more github.com/roboflow/super…

the full tutorial will be available on Monday on the @roboflow YouTube channel; subscribe so you don't miss it.

link to YouTube: youtube.com/@Roboflow

to ensure the privacy of store employees, I decided to blur their faces; to do this, we will first need a model capable of detecting them.

yesterday, I quickly labeled a few dozen images, which I used to train my model.

this is enough for the demo, but we need many more images to implement such a use case.

dataset link: universe.roboflow.com/roboflow-jvuqo…

here's the result! I trained the model in Google Colab.

colab link: colab.research.google.com/github/roboflo…

I used the inference package to run a model pre-trained on the COCO dataset to detect people. I used ByteTrack for tracking.

you can learn more about the Inference + Supervision combo from this tutorial.

tutorial link: supervision.roboflow.com/latest/how_to/…

we want boxes to be stable; to reduce box flickering, we'll use smoothing - averaging the box positions based on the last N frames.

pay attention to the customer #37; on the left without smoothing, on the right with smoothing.

smoothing docs: supervision.roboflow.com/latest/detecti…

I drew the zones using MakeSense - an open-source photo labeling program I created while living in a dormitory several years ago.

make sense Github link: github.com/SkalskiP/make-…

here is a fragment of the logic responsible for counting time.

- if you're working with static video files, I recommend using an FPS-based approach.
- I recommend a clock-based approach if you're working with video streams.

calculating the time customers spend waiting in line at a store is just one of the potential applications.

here's another use case where we calculate how long drivers wait to pass through an intersection.