Model Details
Model Description
A pretrained, general purpose microscopy identity tracking model. Trained on diverse data across scales, from cells to organelles. Expects videos with accompanying detection labels in .slp or CellTrackingChallenge format.
- Developed by: talmolab
- Shared by: talmolab
- License: [More Information Needed]
Model Sources
- Repository: talmolab/dreem
- Paper [optional]: [More Information Needed]
Uses
To track identities of instances in microscopy timelapses. Maintains consistent identities across time.
How to Get Started with the Model
Training Details
Training Data
~100k frames of proofread identity tracked public data. See https://dreem.sleap.ai/latest/datasets
[More Information Needed]
Training Hyperparameters
See configuration files provided in model directory
Metrics
CLEARMOT metrics, implemented by https://github.com/cheind/py-motmetrics
CellTrackingChallenge metrics, implemented by https://github.com/CellTrackingChallenge/py-ctcmetrics#
Hardware
Trained on 4x A40 GPUs
Citation [optional]
BibTeX:
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support