docs(notebooks): 📝 json and csv sink cookbooks are added

[onuralpszr]

trafficstars

Description

New JSON and CSV sink cookbooks are added. For Json it needs a version bump, I used the direct develop branch while preparing it.

JSON collab URL: https://colab.research.google.com/drive/19aqX0QwKXxRk4sYtdEmghO2rzEAKMciM?usp=sharing CSV collab URL: https://colab.research.google.com/drive/1-qetEcyzrBCOCq-TfVVT8AgUG3rJUhT7?usp=sharing

[onuralpszr]

@SkalskiP do you want to merge/review this?

[SkalskiP]

Hi @onuralpszr 👋🏻, thanks for the reminder.

• It would be awesome to add some life to those cookbooks. Display images and/or videos. People see that you are downloading the VideoAssets.PEOPLE_WALKING video, but it would be cool to display it. Take a look here for reference.

• You used supervision==0.19.0rc5 version. Let's bump it to supervision==0.21.0rc5. We rolled out some tracker bugs in supervision==0.20.0.

• That link https://inference.roboflow.com/reference_pages/model_aliases/ is no longer live.

• I'd move this 'detections.csv' to constant.

• You no longer need that. You can use aliases directly.

  REGISTERED_ALIASES = {
      "yolov8n-640": "coco/3",
      "yolov8n-1280": "coco/9",
      "yolov8m-640": "coco/8",
      "yolov8x-1280": "coco/10",
  }
  

• Call byte_track.reset() after this byte_track = sv.ByteTrack(). Tracker ID's in your CSV start at 180.

• Use minimum_consecutive_frames=3 in byte_track = sv.ByteTrack(). It will make results more stable.

[onuralpszr]

@SkalskiP thank you for tips let me update all.

[SkalskiP]

Pleasure 💜

[onuralpszr]

@SkalskiP I updated both collabs in "the links" could you check, after your confirm I will commit all