Update README.md
Browse files
README.md
CHANGED
@@ -4,7 +4,7 @@ pipeline_tag: object-detection
|
|
4 |
tags:
|
5 |
- Pose Estimation
|
6 |
---
|
7 |
-
RTMO / YOLO-NAS-Pose Inference with CUDAExecutionProvider / TensorrtExecutionProvider DEMO
|
8 |
|
9 |
- `demo.sh`: DEMO main program, which will first install rtmlib, and then use rtmo-s to analyze the .mp4 files in the video folder.
|
10 |
- `demo_batch.sh`: Multi-batch version of demo.sh
|
@@ -16,13 +16,15 @@ RTMO / YOLO-NAS-Pose Inference with CUDAExecutionProvider / TensorrtExecutionPro
|
|
16 |
- `rtmo_demo_batch.py`: Multi-batch version of demo_batch.sh
|
17 |
- `video`: Contains one test video.
|
18 |
|
19 |
-
|
|
|
|
|
|
|
|
|
20 |
|
21 |
We did the following to make them work with TensorRTExecutionProvdier
|
22 |
|
23 |
1. Shape inference
|
24 |
2. batch size 1,2,4 fixation
|
25 |
|
26 |
-
|
27 |
-
|
28 |
-
FP16 ONNX model is also provided.
|
|
|
4 |
tags:
|
5 |
- Pose Estimation
|
6 |
---
|
7 |
+
## RTMO / YOLO-NAS-Pose Inference with CUDAExecutionProvider / TensorrtExecutionProvider DEMO
|
8 |
|
9 |
- `demo.sh`: DEMO main program, which will first install rtmlib, and then use rtmo-s to analyze the .mp4 files in the video folder.
|
10 |
- `demo_batch.sh`: Multi-batch version of demo.sh
|
|
|
16 |
- `rtmo_demo_batch.py`: Multi-batch version of demo_batch.sh
|
17 |
- `video`: Contains one test video.
|
18 |
|
19 |
+
# Note
|
20 |
+
|
21 |
+
* Original ONNX models come from [MMPOSE/RTMO Project Page](https://github.com/open-mmlab/mmpose/tree/main/projects/rtmo) trained on body7. We did only
|
22 |
+
* DEMO Inferecne Code is modified from [rtmlib](https://github.com/Tau-J/rtmlib)
|
23 |
+
* TensorrtExecutionProvider only supports Models with fixed batch size (*_batchN.onnx) while CUDAExecutionProvider can run with dynamic batch size.
|
24 |
|
25 |
We did the following to make them work with TensorRTExecutionProvdier
|
26 |
|
27 |
1. Shape inference
|
28 |
2. batch size 1,2,4 fixation
|
29 |
|
30 |
+
PS. FP16 ONNX model is also provided.
|
|
|
|