Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Include DEEP SORT TRACKING #594

Open
1 of 2 tasks
yeongnamtan opened this issue Nov 14, 2023 · 3 comments
Open
1 of 2 tasks

Include DEEP SORT TRACKING #594

yeongnamtan opened this issue Nov 14, 2023 · 3 comments
Assignees
Labels
enhancement New feature or request

Comments

@yeongnamtan
Copy link

Search before asking

  • I have searched the Supervision issues and found no similar feature requests.

Description

Would you consider including DEEP SORT Tracker in Supervision ?

Use case

No response

Additional

No response

Are you willing to submit a PR?

  • Yes I'd like to help by submitting a PR!
@yeongnamtan yeongnamtan added the enhancement New feature or request label Nov 14, 2023
@SkalskiP
Copy link
Collaborator

Hi, @yeongnamtan! 👋🏻 Correct me if I'm wrong, but DeepSort requires loading the model in PyTorch or TensorFlow. We are trying to make Supervision not require installation of such heavy dependencies.

@yeongnamtan
Copy link
Author

@SkalskiP I did a comparison between ByteTrack (using LINE COUNTER), and DeepSort for the same source video. DeepSort performed much better in terms of counting.
Picture1

For ByteTrack, my settings as follow:
byte_tracker = sv.ByteTrack(track_thresh=0.3, track_buffer=60, match_thresh=0.9, frame_rate=30)

For DeepSort, confidence level same at 30%
DEEPSORT:
MODEL_TYPE: "osnet_x_25"
MAX_DIST: 0.1 # The matching threshold. Samples with larger distance are considered an invalid match
MAX_IOU_DISTANCE: 0.7 # Gating threshold. Associations with cost larger than this value are disregarded.
MAX_AGE: 30 # Maximum number of missed misses before a track is deleted
N_INIT: 3 # Number of frames that a track remains in initialization phase
NN_BUDGET: 100 # Maximum size of the appearance descriptors gallery

@tteresi7
Copy link

Hi, @yeongnamtan! 👋🏻 Correct me if I'm wrong, but DeepSort requires loading the model in PyTorch or TensorFlow. We are trying to make Supervision not require installation of such heavy dependencies.

DeepSORT has no such requirements. It only requires the bounding box detection and a feature vector. The feature vector can be generated via a secondary model or extracting a specific layer of another model. It doesn't matter if it's TF, PyTorch or whatever.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

3 participants