- The implementation for "ProFiT: A Prompt-Guided Frequency-Aware Filtering and Template-Enhanced Interaction Framework for Hyperspectral Video Tracking".
- ISPRS Journal of Photogrammetry and Remote Sensing, 2025.
🏃Keep updating🏃:
- Trained model of ProFiT has been released.
- Training and testing code of ProFiT has been released.
- Tracking result of ProFiT has been released.
| Benchmark | ProFiT (Pre/Suc) |
|---|---|
| HOTC20 | 0.971 / 0.758 |
| NIR23 | 0.947 / 0.754 |
| RedNIR23 | 0.755 / 0.613 |
| VIS23 | 0.915 / 0.720 |
| NIR24 | 0.949 / 0.763 |
| RedNIR24 | 0.750 / 0.581 |
| VIS24 | 0.721 / 0.561 |
| MSSOT | 0.857 / 0.602 |
| MSVT | 0.967 / 0.754 |
- Red --> ProFiT. Blue --> Ground Truth. The Rest --> Compared Arts.
git clone https://github.com/YZCU/ProFiT.git
- CUDA 11.8
- Python 3.9.18
- PyTorch 2.0.0
- Torchvision 0.15.0
- numpy 1.25.0
-
Training: Please download the hyperspectral training and testing sets: HOTC20, HOTC23, HOTC24, MSSOT, MSVT.
-
Fast Training: Download the pre-trained model of ProFiT. Put it into
<pretrained_models>. -
Run
<tracking/train.py>to train ProFiT. -
The well-trained ProFiT model is put into
<output/train/yzcu/yzcu-ep150-full-256_shared/yzcu_ep0015.pth.tar>. -
We have also released the well-trained ProFiT tracking models.
-
Testing: Run
<tracking/test.py>for testing, and results are saved in<output/results/yzcu/yzcu-ep150-full-256_shared>. -
Evaluating: Please download the evaluation benchmark Toolkit and vlfeat for more accurate evaluation.
-
Refer to the Hyperspectral Object Tracking Challenge for detailed evaluations.
-
Evaluation of the ProFiT tracker. Run
<tracker_benchmark_v1.0\perfPlot.m>
- If you have any questions or suggestions, feel free to contact me.
- Email: yzchen1006@163.com
❤️ ❤️ We sincerely appreciate the insightful feedback provided by Editors and Reviewers. ❤️ ❤️