CVPR - Prework

[paularamo]

We will start this conversation with a big WELCOME to the CVPR community, we are pleased you have chosen this tutorial. Here, you will learn how to set up and run many AI models with OpenVINO and your webcam in just 10 minutes. As part of the tutorial, you should set up your laptop, in order to run the demonstration we have prepared for you.

• Follow the instructions of this Wiki page.
• Run the notebooks 301 and 401.
• Share a screenshot in this thread.

[pfvideo]

Screenshot of notebooks 301:

Notebooks 401:

[paularamo]

Thanks @pfvideo. Could you please share the results after running the notebooks? 👍

[manduk-ai]

Hi,
A pre-work was a great idea - thanks.
Some of my screenshots.

301 training notebook - inference:

---

301 pot notebook: Accuracy looks the same for the compressed model:

---

301 pot notebook - inference:

---

401 notebook - a "knife" ;-), the algo seems a little bit over aggressive

---

401 notebook - real time detection (camera). Sorry, I'm still a little bit shaken after yesterday's FOMC:

---

401 notebook - also ok with a video as a source

[manduk-ai]

Also: don't have a supported GPU so no data from the very interesting "Compare Inference Speed" part.

[paularamo]

Thanks @AIGeekProgrammer. Well done! See you on Sunday 👍

[mengjian0502]

Hi there,

Nice pre-work! Does the OpenVino supports Pytorch models?

Here're some of my screenshots:

Notebook 301

---

Notebook 301 - POT

---

Notebook 401 - Video-based detection

See you on Sunday!

[paularamo]

Hey, great job, thanks for sharing

[kilianyp]

Worked without any problems 👍 See you tomorrow

301

401

[paularamo]

See you soon, thanks for sharing!

[malamleh93]

301

401

[hcarrillo]

Hi my 401 notebook result

[hcarrillo]

Results of 301 notebook

[raymondlo84]

Neat!

[TanguyJeanneau]

401

301
Not all dataset seems to load for 301 though

[raymondlo84]

Oh! I remember you now! :) Thanks for coming again!

[rvainshtein]

Quick & dirty way to insert inference on a picture of a cat 😉

[paularamo]

Thanks all for your participation in our tutorial!