Hugging Face's logo

qualcomm
/
Track-Anything

PyTorch
android
video-object-tracking
Model card Files
xet
 Community
Track-Anything / README.md
qaihm-bot's picture
qaihm-bot
v0.59.0
738f2b4 verified
preview code
|
Raw
History Blame Contribute Delete
13.6 kB
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: video-object-tracking
---
![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/track_anything/web-assets/model_demo.png)
# Track-Anything: Optimized for Qualcomm Devices
Track-Anything is a video based machine learning model to track an object in a video.
This is based on the implementation of Track-Anything found [here](https://github.com/gaomingqi/Track-Anything).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/track_anything) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
## Getting Started
There are two ways to deploy this model on your device:
### Option 1: Download Pre-Exported Models
Below are pre-exported model assets ready for deployment.
| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/track_anything/releases/v0.59.0/track_anything-onnx-float.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/track_anything/releases/v0.59.0/track_anything-tflite-float.zip)
For more device-specific assets and performance metrics, visit **[Track-Anything on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/track_anything)**.
### Option 2: Export with Custom Configurations
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/track_anything) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations
This option is ideal if you need to customize the model beyond the default configuration provided here.
See our repository for [Track-Anything on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/track_anything) for usage instructions.
## Model Details
**Model Type:** Model_use_case.video_object_tracking
**Model Stats:**
- Model checkpoint: xmem
- Input resolution: 320x568
- Number of parameters (encode_key_with_shrinkage): 9.72M
- Model size (encode_key_with_shrinkage) (float): 37.1 MB
- Number of parameters (encode_value): 23.3M
- Model size (encode_value) (float): 88.8 MB
- Number of parameters (encode_key_without_shrinkage): 9.71M
- Model size (encode_key_without_shrinkage) (float): 37.1 MB
- Number of parameters (segment): 30.1M
- Model size (segment) (float): 115 MB
## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| encode_key_with_shrinkage | ONNX | float | Snapdragon® X2 Elite | 3.939 ms | 10 - 10 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Snapdragon® X Elite | 5.392 ms | 18 - 18 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 3.686 ms | 0 - 198 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 10.648 ms | 9 - 191 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 5.22 ms | 0 - 20 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Qualcomm® QCS8450 | 10.648 ms | 9 - 191 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 8.802 ms | 8 - 11 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 5.392 ms | 18 - 18 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 3.004 ms | 0 - 169 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Snapdragon® 8 Elite Mobile | 3.004 ms | 0 - 169 MB | NPU
| encode_key_with_shrinkage | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 2.377 ms | 0 - 174 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 4.065 ms | 0 - 210 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 11.713 ms | 0 - 194 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 31.245 ms | 3 - 179 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 5.937 ms | 4 - 6 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® SA8775P | 9.457 ms | 3 - 181 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® SA8650P | 9.457 ms | 3 - 181 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® SA8255P | 9.457 ms | 3 - 181 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® QCS8450 | 11.713 ms | 0 - 194 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 9.159 ms | 2 - 29 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 3.293 ms | 0 - 180 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® SA7255P | 31.245 ms | 3 - 179 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Qualcomm® SA8295P | 10.348 ms | 3 - 163 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Snapdragon® 8 Elite Mobile | 3.293 ms | 0 - 180 MB | NPU
| encode_key_with_shrinkage | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 2.467 ms | 0 - 184 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® X2 Elite | 3.886 ms | 80 - 80 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® X Elite | 6.806 ms | 80 - 80 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 5.046 ms | 55 - 257 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 13.545 ms | 44 - 247 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 6.863 ms | 24 - 64 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Qualcomm® QCS8450 | 13.545 ms | 44 - 247 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 10.621 ms | 52 - 55 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 6.806 ms | 80 - 80 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 4.385 ms | 52 - 248 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® 8 Elite Mobile | 4.385 ms | 52 - 248 MB | NPU
| encode_key_without_shrinkage | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.474 ms | 0 - 178 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 5.58 ms | 0 - 214 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 14.156 ms | 0 - 202 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 33.555 ms | 11 - 194 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 7.794 ms | 14 - 16 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® SA8775P | 11.674 ms | 20 - 207 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® SA8650P | 11.674 ms | 20 - 207 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® SA8255P | 11.674 ms | 20 - 207 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® QCS8450 | 14.156 ms | 0 - 202 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 11.337 ms | 20 - 63 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.42 ms | 18 - 203 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® SA7255P | 33.555 ms | 11 - 194 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Qualcomm® SA8295P | 12.718 ms | 20 - 184 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Snapdragon® 8 Elite Mobile | 4.42 ms | 18 - 203 MB | NPU
| encode_key_without_shrinkage | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.574 ms | 0 - 221 MB | NPU
| encode_value | ONNX | float | Snapdragon® X2 Elite | 6.553 ms | 20 - 20 MB | NPU
| encode_value | ONNX | float | Snapdragon® X Elite | 13.249 ms | 39 - 39 MB | NPU
| encode_value | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 11.379 ms | 0 - 209 MB | NPU
| encode_value | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 31.596 ms | 12 - 208 MB | NPU
| encode_value | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 16.791 ms | 0 - 50 MB | NPU
| encode_value | ONNX | float | Qualcomm® QCS8450 | 31.596 ms | 12 - 208 MB | NPU
| encode_value | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 20.338 ms | 6 - 14 MB | NPU
| encode_value | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 13.249 ms | 39 - 39 MB | NPU
| encode_value | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 9.547 ms | 7 - 191 MB | NPU
| encode_value | ONNX | float | Snapdragon® 8 Elite Mobile | 9.547 ms | 7 - 191 MB | NPU
| encode_value | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.222 ms | 5 - 179 MB | NPU
| encode_value | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 8.241 ms | 1 - 286 MB | NPU
| encode_value | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 21.253 ms | 3 - 291 MB | NPU
| encode_value | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 51.329 ms | 3 - 217 MB | NPU
| encode_value | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 11.857 ms | 3 - 6 MB | NPU
| encode_value | TFLITE | float | Qualcomm® SA8775P | 17.008 ms | 3 - 215 MB | NPU
| encode_value | TFLITE | float | Qualcomm® SA8650P | 17.008 ms | 3 - 215 MB | NPU
| encode_value | TFLITE | float | Qualcomm® SA8255P | 17.008 ms | 3 - 215 MB | NPU
| encode_value | TFLITE | float | Qualcomm® QCS8450 | 21.253 ms | 3 - 291 MB | NPU
| encode_value | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 17.086 ms | 0 - 55 MB | NPU
| encode_value | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 6.861 ms | 3 - 213 MB | NPU
| encode_value | TFLITE | float | Qualcomm® SA7255P | 51.329 ms | 3 - 217 MB | NPU
| encode_value | TFLITE | float | Qualcomm® SA8295P | 18.842 ms | 0 - 216 MB | NPU
| encode_value | TFLITE | float | Snapdragon® 8 Elite Mobile | 6.861 ms | 3 - 213 MB | NPU
| encode_value | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.401 ms | 1 - 201 MB | NPU
| segment | ONNX | float | Snapdragon® X2 Elite | 11.676 ms | 21 - 21 MB | NPU
| segment | ONNX | float | Snapdragon® X Elite | 23.175 ms | 37 - 37 MB | NPU
| segment | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 17.336 ms | 25 - 267 MB | NPU
| segment | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 43.443 ms | 2 - 232 MB | NPU
| segment | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 23.068 ms | 0 - 45 MB | NPU
| segment | ONNX | float | Qualcomm® QCS8450 | 43.443 ms | 2 - 232 MB | NPU
| segment | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 35.156 ms | 21 - 45 MB | NPU
| segment | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 23.175 ms | 37 - 37 MB | NPU
| segment | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 13.315 ms | 5 - 211 MB | NPU
| segment | ONNX | float | Snapdragon® 8 Elite Mobile | 13.315 ms | 5 - 211 MB | NPU
| segment | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 10.822 ms | 5 - 216 MB | NPU
| segment | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 17.433 ms | 1 - 292 MB | NPU
| segment | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 42.34 ms | 2 - 294 MB | NPU
| segment | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 103.337 ms | 2 - 195 MB | NPU
| segment | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 23.541 ms | 2 - 4 MB | NPU
| segment | TFLITE | float | Qualcomm® SA8775P | 33.286 ms | 2 - 193 MB | NPU
| segment | TFLITE | float | Qualcomm® SA8650P | 33.286 ms | 2 - 193 MB | NPU
| segment | TFLITE | float | Qualcomm® SA8255P | 33.286 ms | 2 - 193 MB | NPU
| segment | TFLITE | float | Qualcomm® QCS8450 | 42.34 ms | 2 - 294 MB | NPU
| segment | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 35.307 ms | 0 - 83 MB | NPU
| segment | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 13.196 ms | 0 - 195 MB | NPU
| segment | TFLITE | float | Qualcomm® SA7255P | 103.337 ms | 2 - 195 MB | NPU
| segment | TFLITE | float | Qualcomm® SA8295P | 37.583 ms | 2 - 202 MB | NPU
| segment | TFLITE | float | Snapdragon® 8 Elite Mobile | 13.196 ms | 0 - 195 MB | NPU
| segment | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 11.574 ms | 2 - 197 MB | NPU
## License
* The license for the original implementation of Track-Anything can be found
 [here](https://github.com/gaomingqi/Track-Anything/blob/master/LICENSE).
## References
* [Track Anything: Segment Anything Meets Videos](https://arxiv.org/abs/2304.11968)
* [Source Model Implementation](https://github.com/gaomingqi/Track-Anything)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).