OpenMMLab
/
mmpose
mirror of https://github.com/open-mmlab/mmpose
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
mmpose/projects/rtmpose/benchmark/README_CN.md

7.9 KiB
Raw Permalink Blame History

RTMPose Benchmarks

简体中文 | English

欢迎社区用户在不同硬件设备上进行推理速度测试，贡献到本项目目录下。

当前已测试：

  • CPU
    • Intel i7-11700
  • GPU
    • NVIDIA GeForce 1660 Ti
    • NVIDIA GeForce RTX 3090
  • Nvidia Jetson
    • AGX Orin
    • Orin NX
  • ARM
    • Snapdragon 865

人体 2d 关键点 (17 Keypoints)

Model Info

Config Input Size AP
(COCO)		 Params(M) FLOPS(G)
RTMPose-t 256x192 68.5 3.34 0.36
RTMPose-s 256x192 72.2 5.47 0.68
RTMPose-m 256x192 75.8 13.59 1.93
RTMPose-l 256x192 76.5 27.66 4.16
RTMPose-m 384x288 77.0 13.72 4.33
RTMPose-l 384x288 77.3 27.79 9.35

Speed Benchmark

图中所示为模型推理时间，单位毫秒。

Config Input Size ORT
(i7-11700)		 TRT-FP16
(GTX 1660Ti)		 TRT-FP16
(RTX 3090)		 ncnn-FP16
(Snapdragon 865)		 TRT-FP16
(Jetson AGX Orin)		 TRT-FP16
(Jetson Orin NX)
RTMPose-t 256x192 3.20 1.06 0.98 9.02 1.63 1.97
RTMPose-s 256x192 4.48 1.39 1.12 13.89 1.85 2.18
RTMPose-m 256x192 11.06 2.29 1.18 26.44 2.72 3.35
RTMPose-l 256x192 18.85 3.46 1.37 45.37 3.67 4.78
RTMPose-m 384x288 24.78 3.66 1.20 26.44 3.45 5.08
RTMPose-l 384x288 - 6.05 1.74 - 4.93 7.23

人体全身 2d 关键点 (133 Keypoints)

Model Info

Config Input Size Whole AP Whole AR FLOPS(G)
RTMPose-m 256x192 60.4 66.7 2.22
RTMPose-l 256x192 63.2 69.4 4.52
RTMPose-l 384x288 67.0 72.3 10.07

Speed Benchmark

  • 图中所示为模型推理时间，单位毫秒。
  • 来自不同社区用户的测试数据用 | 分隔开。
Config Input Size ORT
(i7-11700)		 TRT-FP16
(GTX 1660Ti)		 TRT-FP16
(RTX 3090)		 TRT-FP16
(Jetson AGX Orin)		 TRT-FP16
(Jetson Orin NX)
RTMPose-m 256x192 13.50 4.00 1.17 | 1.84 2.79 3.51
RTMPose-l 256x192 23.41 5.67 1.44 | 2.61 3.80 4.95
RTMPose-l 384x288 44.58 7.68 1.75 | 4.24 5.08 7.20

如何测试推理速度

我们使用 MMDeploy 提供的 tools/profiler.py 脚本进行模型测速。

用户需要准备一个存放测试图片的文件夹./test_images，profiler 将随机从该目录下抽取图片用于模型测速。

python tools/profiler.py \
    configs/mmpose/pose-detection_simcc_onnxruntime_dynamic.py \
    {RTMPOSE_PROJECT}/rtmpose/body_2d_keypoint/rtmpose-m_8xb256-420e_coco-256x192.py \
    ../test_images \
    --model {WORK_DIR}/end2end.onnx \
    --shape 256x192 \
    --device cpu \
    --warmup 50 \
    --num-iter 200

The result is as follows:

01/30 15:06:35 - mmengine - INFO - [onnxruntime]-70 times per count: 8.73 ms, 114.50 FPS
01/30 15:06:36 - mmengine - INFO - [onnxruntime]-90 times per count: 9.05 ms, 110.48 FPS
01/30 15:06:37 - mmengine - INFO - [onnxruntime]-110 times per count: 9.87 ms, 101.32 FPS
01/30 15:06:37 - mmengine - INFO - [onnxruntime]-130 times per count: 9.99 ms, 100.10 FPS
01/30 15:06:38 - mmengine - INFO - [onnxruntime]-150 times per count: 10.39 ms, 96.29 FPS
01/30 15:06:39 - mmengine - INFO - [onnxruntime]-170 times per count: 10.77 ms, 92.86 FPS
01/30 15:06:40 - mmengine - INFO - [onnxruntime]-190 times per count: 10.98 ms, 91.05 FPS
01/30 15:06:40 - mmengine - INFO - [onnxruntime]-210 times per count: 11.19 ms, 89.33 FPS
01/30 15:06:41 - mmengine - INFO - [onnxruntime]-230 times per count: 11.16 ms, 89.58 FPS
01/30 15:06:42 - mmengine - INFO - [onnxruntime]-250 times per count: 11.06 ms, 90.41 FPS
----- Settings:
+------------+---------+
| batch size |    1    |
|   shape    | 256x192 |
| iterations |   200   |
|   warmup   |    50   |
+------------+---------+
----- Results:
+--------+------------+---------+
| Stats  | Latency/ms |   FPS   |
+--------+------------+---------+
|  Mean  |   11.060   |  90.412 |
| Median |   11.852   |  84.375 |
|  Min   |   7.812    | 128.007 |
|  Max   |   13.690   |  73.044 |
+--------+------------+---------+

If you want to learn more details of profiler, you can refer to the Profiler Docs.