MixFormerV2 is an improved unified Transformer-based object tracking model that enhances tracking accuracy and speed by integrating the feature extraction and matching processes.
git clone https://huggingface.co/AXERA-TECH/MixFormerV2
File Description:
m5stack@raspberrypi:~/rsp/MixFormerV2 $ ls -lh
total 63M
drwxrwxr-x 2 m5stack m5stack 4.0K Aug 11 18:28 ax630c
drwxrwxr-x 2 m5stack m5stack 4.0K Aug 11 18:28 ax650
-rw-rw-r-- 1 m5stack m5stack 63M Aug 11 18:28 car.avi
-rw-rw-r-- 1 m5stack m5stack 0 Aug 11 18:28 config.json
drwxrwxr-x 2 m5stack m5stack 4.0K Aug 11 18:28 onnx
-rw-rw-r-- 1 m5stack m5stack 4.0K Aug 11 18:28 README.md
-rw-rw-r-- 1 m5stack m5stack 15K Aug 11 18:28 run_mixformer2_axmodel.py
-rw-rw-r-- 1 m5stack m5stack 14K Aug 11 18:28 run_mixformer2_onnx.py
python -m venv mixformer
source mixformer/bin/activate
pip install https://github.com/AXERA-TECH/pyaxengine/releases/download/0.1.3.rc1/axengine-0.1.3-py3-none-any.whl
pip install argparse numpy opencv-python glob2
python3 run_mixformer2_axmodel.py --model-path ax650/mixformer_v2.axmodel --frame-path car.avi -r 10
Example Output:
(mixformer) m5stack@raspberrypi:~/rsp/MixFormerV2 $ python3 run_mixformer2_axmodel.py --model-path ax650/mixformer_v2.axmodel --frame-path car.avi -r 10
[INFO] Available providers: ['AXCLRTExecutionProvider']
[INFO] Using provider: AXCLRTExecutionProvider
[INFO] SOC Name: AX650N
[INFO] VNPU type: VNPUType.DISABLED
[INFO] Compiler version: 3.4-dirty 4ff37520-dirty
====================type================= [1079, 482] <class 'list'> <class 'list'>
First frame initialized!
Video: tracking 1011.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Video: tracking 8.0fps
Reached the maximum number of frames (10). Exiting loop.
video: average finale average tracking fps 121.2 fps