From c8c48e4bf23e1f8feb8cdd1c88b1f35423435604 Mon Sep 17 00:00:00 2001
From: Daniel <daniel@mail.de>
Date: Mon, 2 Dec 2024 13:50:35 +0100
Subject: [PATCH] Using TensorRT backend.

---
 media/RESULTS.md             | 282 +++++++++++++++++------------------
 scripts/utils_2d_pose_ort.py |  13 +-
 2 files changed, 149 insertions(+), 146 deletions(-)

diff --git a/media/RESULTS.md b/media/RESULTS.md
index af23192..0e7e5ae 100644
--- a/media/RESULTS.md
+++ b/media/RESULTS.md
@@ -6,9 +6,9 @@ Results of the model in various experiments on different datasets.
 
 ```json
 {
-  "avg_time_2d": 0.027316828905525854,
-  "avg_time_3d": 0.0004260645074359441,
-  "avg_fps": 36.04526698476189
+  "avg_time_2d": 0.019079747846571064,
+  "avg_time_3d": 0.0003823995590209961,
+  "avg_fps": 51.381791492991674
 }
 {
   "person_nums": {
@@ -27,248 +27,248 @@ Results of the model in various experiments on different datasets.
   },
   "mpjpe": {
     "count": 600,
-    "mean": 0.067549,
-    "median": 0.059367,
-    "std": 0.027836,
+    "mean": 0.067471,
+    "median": 0.059656,
+    "std": 0.027837,
     "sem": 0.001137,
-    "min": 0.044839,
-    "max": 0.190705,
+    "min": 0.04161,
+    "max": 0.191019,
     "recall-0.025": 0.0,
-    "recall-0.05": 0.046667,
+    "recall-0.05": 0.045,
     "recall-0.1": 0.93,
     "recall-0.15": 0.95,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600,
     "ap-0.025": 0.0,
-    "ap-0.05": 0.004947,
-    "ap-0.1": 0.890901,
-    "ap-0.15": 0.915998,
+    "ap-0.05": 0.005008,
+    "ap-0.1": 0.890421,
+    "ap-0.15": 0.915856,
     "ap-0.25": 1.0,
     "ap-0.5": 1.0
   },
   "nose": {
     "count": 600,
-    "mean": 0.116488,
-    "median": 0.100674,
-    "std": 0.042049,
-    "sem": 0.001718,
-    "min": 0.037604,
-    "max": 0.277341,
-    "recall-0.025": 0.0,
-    "recall-0.05": 0.008333,
-    "recall-0.1": 0.493333,
-    "recall-0.15": 0.82,
-    "recall-0.25": 0.991667,
+    "mean": 0.116291,
+    "median": 0.101326,
+    "std": 0.04236,
+    "sem": 0.001731,
+    "min": 0.021073,
+    "max": 0.288626,
+    "recall-0.025": 0.001667,
+    "recall-0.05": 0.006667,
+    "recall-0.1": 0.491667,
+    "recall-0.15": 0.821667,
+    "recall-0.25": 0.99,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "shoulder_left": {
     "count": 600,
-    "mean": 0.03369,
-    "median": 0.025989,
-    "std": 0.031823,
+    "mean": 0.033847,
+    "median": 0.025987,
+    "std": 0.031824,
     "sem": 0.0013,
-    "min": 0.001966,
-    "max": 0.182109,
-    "recall-0.025": 0.476667,
+    "min": 0.002728,
+    "max": 0.181894,
+    "recall-0.025": 0.466667,
     "recall-0.05": 0.866667,
     "recall-0.1": 0.948333,
-    "recall-0.15": 0.966667,
+    "recall-0.15": 0.968333,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "shoulder_right": {
     "count": 600,
-    "mean": 0.048891,
-    "median": 0.034053,
-    "std": 0.042652,
-    "sem": 0.001743,
-    "min": 0.004714,
-    "max": 0.251305,
-    "recall-0.025": 0.238333,
-    "recall-0.05": 0.756667,
-    "recall-0.1": 0.901667,
-    "recall-0.15": 0.943333,
+    "mean": 0.04973,
+    "median": 0.034604,
+    "std": 0.044052,
+    "sem": 0.0018,
+    "min": 0.00395,
+    "max": 0.256048,
+    "recall-0.025": 0.211667,
+    "recall-0.05": 0.753333,
+    "recall-0.1": 0.906667,
+    "recall-0.15": 0.936667,
     "recall-0.25": 0.998333,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "elbow_left": {
     "count": 600,
-    "mean": 0.043307,
-    "median": 0.034999,
-    "std": 0.034727,
-    "sem": 0.001419,
-    "min": 0.001548,
-    "max": 0.196855,
-    "recall-0.025": 0.253333,
-    "recall-0.05": 0.808333,
-    "recall-0.1": 0.945,
-    "recall-0.15": 0.953333,
+    "mean": 0.042919,
+    "median": 0.035054,
+    "std": 0.034438,
+    "sem": 0.001407,
+    "min": 0.00136,
+    "max": 0.198368,
+    "recall-0.025": 0.246667,
+    "recall-0.05": 0.825,
+    "recall-0.1": 0.948333,
+    "recall-0.15": 0.955,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "elbow_right": {
     "count": 600,
-    "mean": 0.043417,
-    "median": 0.032847,
-    "std": 0.03524,
-    "sem": 0.00144,
-    "min": 0.006341,
-    "max": 0.220952,
-    "recall-0.025": 0.246667,
-    "recall-0.05": 0.813333,
-    "recall-0.1": 0.93,
-    "recall-0.15": 0.941667,
-    "recall-0.25": 1.0,
+    "mean": 0.043901,
+    "median": 0.033077,
+    "std": 0.036324,
+    "sem": 0.001484,
+    "min": 0.006465,
+    "max": 0.274304,
+    "recall-0.025": 0.241667,
+    "recall-0.05": 0.81,
+    "recall-0.1": 0.925,
+    "recall-0.15": 0.943333,
+    "recall-0.25": 0.998333,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "wrist_left": {
     "count": 600,
-    "mean": 0.042168,
-    "median": 0.026768,
-    "std": 0.042016,
-    "sem": 0.001717,
-    "min": 0.002021,
-    "max": 0.200389,
-    "recall-0.025": 0.473333,
-    "recall-0.05": 0.733333,
-    "recall-0.1": 0.915,
-    "recall-0.15": 0.945,
+    "mean": 0.043044,
+    "median": 0.026871,
+    "std": 0.04257,
+    "sem": 0.001739,
+    "min": 0.000214,
+    "max": 0.202968,
+    "recall-0.025": 0.466667,
+    "recall-0.05": 0.731667,
+    "recall-0.1": 0.903333,
+    "recall-0.15": 0.941667,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "wrist_right": {
     "count": 600,
-    "mean": 0.044909,
-    "median": 0.026831,
-    "std": 0.055206,
-    "sem": 0.002256,
-    "min": 0.003601,
-    "max": 0.476436,
-    "recall-0.025": 0.453333,
-    "recall-0.05": 0.783333,
-    "recall-0.1": 0.896667,
-    "recall-0.15": 0.913333,
-    "recall-0.25": 0.993333,
+    "mean": 0.044102,
+    "median": 0.026291,
+    "std": 0.050657,
+    "sem": 0.00207,
+    "min": 0.003409,
+    "max": 0.460683,
+    "recall-0.025": 0.466667,
+    "recall-0.05": 0.776667,
+    "recall-0.1": 0.895,
+    "recall-0.15": 0.915,
+    "recall-0.25": 0.996667,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "hip_left": {
     "count": 600,
-    "mean": 0.090573,
-    "median": 0.087143,
-    "std": 0.032802,
+    "mean": 0.090158,
+    "median": 0.08564,
+    "std": 0.032791,
     "sem": 0.00134,
-    "min": 0.019451,
-    "max": 0.236078,
+    "min": 0.014562,
+    "max": 0.236403,
     "recall-0.025": 0.001667,
-    "recall-0.05": 0.033333,
-    "recall-0.1": 0.798333,
-    "recall-0.15": 0.946667,
+    "recall-0.05": 0.036667,
+    "recall-0.1": 0.82,
+    "recall-0.15": 0.945,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "hip_right": {
     "count": 600,
-    "mean": 0.114861,
-    "median": 0.115523,
-    "std": 0.02614,
-    "sem": 0.001068,
-    "min": 0.044367,
-    "max": 0.234619,
+    "mean": 0.11407,
+    "median": 0.114481,
+    "std": 0.02606,
+    "sem": 0.001065,
+    "min": 0.045083,
+    "max": 0.233275,
     "recall-0.025": 0.0,
-    "recall-0.05": 0.003333,
-    "recall-0.1": 0.228333,
-    "recall-0.15": 0.946667,
+    "recall-0.05": 0.001667,
+    "recall-0.1": 0.251667,
+    "recall-0.15": 0.948333,
     "recall-0.25": 1.0,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "knee_left": {
     "count": 600,
-    "mean": 0.061349,
-    "median": 0.044913,
-    "std": 0.061365,
-    "sem": 0.002507,
-    "min": 0.019282,
-    "max": 0.432248,
-    "recall-0.025": 0.046667,
-    "recall-0.05": 0.598333,
-    "recall-0.1": 0.913333,
+    "mean": 0.061433,
+    "median": 0.045616,
+    "std": 0.060807,
+    "sem": 0.002484,
+    "min": 0.018825,
+    "max": 0.425003,
+    "recall-0.025": 0.058333,
+    "recall-0.05": 0.593333,
+    "recall-0.1": 0.915,
     "recall-0.15": 0.923333,
-    "recall-0.25": 0.978333,
+    "recall-0.25": 0.98,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "knee_right": {
     "count": 600,
-    "mean": 0.051142,
-    "median": 0.042244,
-    "std": 0.036288,
-    "sem": 0.001483,
-    "min": 0.01554,
-    "max": 0.258279,
-    "recall-0.025": 0.04,
-    "recall-0.05": 0.748333,
+    "mean": 0.050724,
+    "median": 0.042002,
+    "std": 0.036168,
+    "sem": 0.001478,
+    "min": 0.016654,
+    "max": 0.257622,
+    "recall-0.025": 0.038333,
+    "recall-0.05": 0.76,
     "recall-0.1": 0.941667,
     "recall-0.15": 0.945,
-    "recall-0.25": 0.996667,
+    "recall-0.25": 0.998333,
     "recall-0.5": 1.0,
     "num_labels": 600
   },
   "ankle_left": {
     "count": 599,
-    "mean": 0.098762,
-    "median": 0.086486,
-    "std": 0.050411,
-    "sem": 0.002061,
-    "min": 0.036478,
-    "max": 0.490644,
+    "mean": 0.09891,
+    "median": 0.086872,
+    "std": 0.051463,
+    "sem": 0.002104,
+    "min": 0.035674,
+    "max": 0.490677,
     "recall-0.025": 0.0,
-    "recall-0.05": 0.001667,
+    "recall-0.05": 0.003333,
     "recall-0.1": 0.831667,
-    "recall-0.15": 0.938333,
+    "recall-0.15": 0.936667,
     "recall-0.25": 0.98,
     "recall-0.5": 0.998333,
     "num_labels": 600
   },
   "ankle_right": {
     "count": 598,
-    "mean": 0.085796,
-    "median": 0.069833,
-    "std": 0.056865,
-    "sem": 0.002327,
-    "min": 0.02542,
-    "max": 0.440686,
+    "mean": 0.085207,
+    "median": 0.069747,
+    "std": 0.054773,
+    "sem": 0.002242,
+    "min": 0.030417,
+    "max": 0.440549,
     "recall-0.025": 0.0,
-    "recall-0.05": 0.018333,
-    "recall-0.1": 0.878333,
+    "recall-0.05": 0.021667,
+    "recall-0.1": 0.88,
     "recall-0.15": 0.906667,
-    "recall-0.25": 0.973333,
+    "recall-0.25": 0.978333,
     "recall-0.5": 0.996667,
     "num_labels": 600
   },
   "joint_recalls": {
     "num_labels": 7800,
-    "recall-0.025": 0.17103,
-    "recall-0.05": 0.47359,
-    "recall-0.1": 0.81628,
+    "recall-0.025": 0.16897,
+    "recall-0.05": 0.47551,
+    "recall-0.1": 0.81936,
     "recall-0.15": 0.92923,
-    "recall-0.25": 0.99269,
+    "recall-0.25": 0.99333,
     "recall-0.5": 0.99949
   }
 }
 {
   "total_parts": 8400,
-  "correct_parts": 8092,
-  "pcp": 0.963333
+  "correct_parts": 8091,
+  "pcp": 0.963214
 }
 ```
 
diff --git a/scripts/utils_2d_pose_ort.py b/scripts/utils_2d_pose_ort.py
index 7d5df48..d48bed4 100644
--- a/scripts/utils_2d_pose_ort.py
+++ b/scripts/utils_2d_pose_ort.py
@@ -13,9 +13,12 @@ class BaseModel(ABC):
     def __init__(self, model_path: str, warmup: int):
         self.opt = ort.SessionOptions()
         providers = ort.get_available_providers()
+        # ort.set_default_logger_severity(1)
 
         provider = ""
-        if "CUDAExecutionProvider" in providers:
+        if "TensorrtExecutionProvider" in providers:
+            provider = "TensorrtExecutionProvider"
+        elif "CUDAExecutionProvider" in providers:
             provider = "CUDAExecutionProvider"
         else:
             provider = "CPUExecutionProvider"
@@ -328,10 +331,10 @@ def load_model():
     print("Loading onnx model ...")
 
     model = TopDown(
-        "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmdet-nano_320x320_with-norm.onnx",
-        # "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmdet-nano_320x320_fp16_with-norm.onnx",
-        "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmpose-m_384x288_with-norm.onnx",
-        # "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmpose-m_384x288_fp16_with-norm.onnx",
+        # "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmdet-nano_320x320_with-norm.onnx",
+        "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmdet-nano_320x320_fp16_with-norm.onnx",
+        # "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmpose-m_384x288_with-norm.onnx",
+        "/RapidPoseTriangulation/extras/mmdeploy/exports/rtmpose-m_384x288_fp16_with-norm.onnx",
         conf_threshold=0.3,
         iou_threshold=0.3,
         warmup=30,