A Coding Guide to Markerless 3D Human Kinematics with Pose2Sim, RTMPose, and OpenSim

import subprocess, sys, os


def run(cmd, desc=""):
   """Run a shell command with live output."""
   print(f"n{'='*60}")
   print(f"  {desc}" if desc else f"  Running: {cmd}")
   print(f"{'='*60}")
   result = subprocess.run(cmd, shell=True, capture_output=True, text=True)
   if result.stdout:
       lines = result.stdout.strip().split('n')
       for line in lines[-15:]:
           print(line)
   if result.returncode != 0:
       print(f" Warning (exit code {result.returncode}): {result.stderr[-500:] if result.stderr else 'unknown error'}")
   return result.returncode


run("pip install -q pose2sim", "Installing Pose2Sim (includes RTMPose, filtering, etc.)")


opensim_available = False
try:
   run("pip install -q opensim", "Installing OpenSim Python bindings")
   import opensim
   opensim_available = True
   print(f" OpenSim {opensim.__version__} installed successfully!")
except Exception as e:
   print(f" OpenSim could not be installed ({e}).")
   print("  The tutorial will run all steps EXCEPT kinematics.")
   print("  For full kinematics, use a local conda environment instead.")


gpu_returncode = run("nvidia-smi --query-gpu=name,memory.total --format=csv,noheader", "Checking GPU availability")
if gpu_returncode != 0:
   print(" No GPU detected. Pose estimation will run on CPU (slower).")
   print("  Tip: Runtime → Change runtime type → T4 GPU")
else:
   print(" GPU detected! Pose estimation will be accelerated.")


try:
   import torch
   print(f" PyTorch {torch.__version__} | CUDA available: {torch.cuda.is_available()}")
except ImportError:
   print(" PyTorch not found (pose estimation will use ONNX Runtime directly)")


print("n Installation complete!")


import shutil
from pathlib import Path


import Pose2Sim
pkg_path = Path(Pose2Sim.__file__).parent
print(f" Pose2Sim installed at: {pkg_path}")


demo_src = pkg_path / "Demo_SinglePerson"
work_dir = Path("/content/Pose2Sim_Tutorial")


if work_dir.exists():
   shutil.rmtree(work_dir)


shutil.copytree(demo_src, work_dir)
print(f" Demo copied to: {work_dir}")


print("n  PROJECT STRUCTURE:")
print("=" * 60)
for p in sorted(work_dir.rglob("*")):
   depth = len(p.relative_to(work_dir).parts)
   indent = "  " * depth
   if p.is_dir():
       print(f"{indent} {p.name}/")
   elif depth <= 3:
       size_kb = p.stat().st_size / 1024
       print(f"{indent} {p.name} ({size_kb:.1f} KB)")


print("""
╔══════════════════════════════════════════════════════════╗
║  KEY DIRECTORIES:                                       ║
║  calibration/  → Camera calibration files               ║
║  videos/       → Raw input videos from each camera      ║
║  pose/         → 2D pose estimation results (auto-gen)  ║
║  pose-3d/      → 3D triangulated coordinates (.trc)     ║
║  kinematics/   → OpenSim joint angles (.mot)            ║
║  Config.toml   → ALL pipeline parameters                ║
╚══════════════════════════════════════════════════════════╝
""")


import toml


config_path = work_dir / "Config.toml"
config = toml.load(config_path)


print(" KEY CONFIGURATION PARAMETERS:")
print("=" * 60)


sections_of_interest = ['project', 'pose', 'calibration', 'synchronization',
                       'triangulation', 'filtering', 'markerAugmentation']
for section in sections_of_interest:
   if section in config:
       print(f"n[{section}]")
       for key, val in list(config[section].items())[:6]:
           print(f"  {key} = {val}")
       if len(config[section]) > 6:
           print(f"  ... and {len(config[section]) - 6} more parameters")


print("nn ADAPTING CONFIG FOR GOOGLE COLAB (headless)...")
print("-" * 60)


modifications = {}


if 'synchronization' in config:
   config['synchronization']['synchronization_gui'] = False
   modifications['synchronization.synchronization_gui'] = False


if 'pose' in config:
   config['pose']['display_detection'] = False
   modifications['pose.display_detection'] = False
   config['pose']['mode'] = 'balanced'
   modifications['pose.mode'] = 'balanced'
   config['pose']['save_video'] = 'none'
   modifications['pose.save_video'] = 'none'
   config['pose']['det_frequency'] = 50
   modifications['pose.det_frequency'] = 50


if 'filtering' in config:
   config['filtering']['display_figures'] = False
   modifications['filtering.display_figures'] = False


if 'triangulation' in config:
   config['triangulation']['undistort_points'] = False
   modifications['triangulation.undistort_points'] = False
   config['triangulation']['handle_LR_swap'] = False
   modifications['triangulation.handle_LR_swap'] = False


if 'kinematics' in config:
   config['kinematics']['use_simple_model'] = True
   modifications['kinematics.use_simple_model'] = True


with open(config_path, 'w') as f:
   toml.dump(config, f)


for k, v in modifications.items():
   print(f"   {k} → {v}")


print("n Config.toml updated for Colab execution!")

import os
import time
import warnings
import matplotlib
matplotlib.use('Agg')


os.chdir(work_dir)
print(f" Working directory: {os.getcwd()}")


from Pose2Sim import Pose2Sim


step_times = {}


print("n" + "█" * 60)
print("█  STEP 1/8: CAMERA CALIBRATION")
print("█" * 60)
print("""
 Calibration converts camera parameters (intrinsics + extrinsics)
 into a unified Pose2Sim format.


 The demo uses 'convert' mode (pre-existing Qualisys calibration).
 For your own data, you can:
   - Convert from Qualisys, Vicon, OpenCap, AniPose, etc.
   - Calculate from scratch using a checkerboard
""")


t0 = time.time()
try:
   Pose2Sim.calibration()
   step_times['calibration'] = time.time() - t0
   print(f"n Calibration complete! ({step_times['calibration']:.1f}s)")


   calib_file = work_dir / "calibration" / "Calib.toml"
   if calib_file.exists():
       calib_data = toml.load(calib_file)
       cam_names = [k for k in calib_data.keys() if k.startswith('cam')]
       print(f"   Cameras found: {len(cam_names)} → {cam_names}")
       for cam in cam_names[:2]:
           print(f"  Camera '{cam}':")
           if 'matrix' in calib_data[cam]:
               print(f"    Intrinsic matrix (3x3): ✓")
           if 'rotation' in calib_data[cam]:
               print(f"    Rotation vector: {calib_data[cam]['rotation']}")
           if 'translation' in calib_data[cam]:
               print(f"    Translation (m): {calib_data[cam]['translation']}")
except Exception as e:
   print(f" Calibration warning: {e}")
   step_times['calibration'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 2/8: 2D POSE ESTIMATION (RTMPose)")
print("█" * 60)
print("""
 RTMPose detects 2D body keypoints in each video frame.
 - Model: RTMPose (balanced mode) with YOLOX detection
 - Body model: Body_with_feet (HALPE_26 keypoints)
 - Output: JSON files per frame, per camera
 - GPU acceleration used if available


 For custom models (animal, hand, face), set `mode` in Config.toml
 to a dictionary with custom model URLs (see README).
""")


t0 = time.time()
try:
   Pose2Sim.poseEstimation()
   step_times['poseEstimation'] = time.time() - t0
   print(f"n Pose estimation complete! ({step_times['poseEstimation']:.1f}s)")


   pose_dirs = list((work_dir / "pose").glob("*_json"))
   for pd in pose_dirs:
       n_files = len(list(pd.glob("*.json")))
       print(f"   {pd.name}: {n_files} JSON frames generated")
except Exception as e:
   print(f" Pose estimation error: {e}")
   step_times['poseEstimation'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 3/8: CAMERA SYNCHRONIZATION")
print("█" * 60)
print("""
 Synchronization aligns frames across cameras by correlating
 vertical keypoint speeds. This finds the time offset between cameras.


 Skip this step if your cameras are hardware-synchronized
 (e.g., via timecode, trigger cable, or GoPro GPS sync).
""")


t0 = time.time()
try:
   Pose2Sim.synchronization()
   step_times['synchronization'] = time.time() - t0
   print(f"n Synchronization complete! ({step_times['synchronization']:.1f}s)")
except Exception as e:
   print(f" Synchronization note: {e}")
   step_times['synchronization'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 4/8: PERSON ASSOCIATION")
print("█" * 60)
print("""
 Associates detected persons across different camera views.
 - Single-person mode: picks person with lowest reprojection error
 - Multi-person mode: uses epipolar geometry to match people


 For this single-person demo, it selects the best-matching person.
""")


t0 = time.time()
try:
   Pose2Sim.personAssociation()
   step_times['personAssociation'] = time.time() - t0
   print(f"n Person association complete! ({step_times['personAssociation']:.1f}s)")
except Exception as e:
   print(f" Person association note: {e}")
   step_times['personAssociation'] = time.time() - t0

print("n" + "█" * 60)
print("█  STEP 5/8: 3D TRIANGULATION")
print("█" * 60)
print("""
 Triangulates 2D keypoints from multiple cameras into 3D coordinates.
 - Weighted by detection confidence scores
 - Cameras removed if reprojection error exceeds threshold
 - Right/left swap correction (disabled per current recommendation)
 - Missing values interpolated (cubic, bezier, linear, etc.)
 - Output: .trc file (OpenSim-compatible 3D marker format)
""")


t0 = time.time()
try:
   Pose2Sim.triangulation()
   step_times['triangulation'] = time.time() - t0
   print(f"n Triangulation complete! ({step_times['triangulation']:.1f}s)")


   trc_files = list((work_dir / "pose-3d").glob("*.trc"))
   for trc in trc_files:
       size_kb = trc.stat().st_size / 1024
       print(f"   {trc.name} ({size_kb:.1f} KB)")
except Exception as e:
   print(f" Triangulation error: {e}")
   step_times['triangulation'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 6/8: 3D COORDINATE FILTERING")
print("█" * 60)
print("""
 Smooths 3D trajectories to reduce noise from triangulation.
 Available filters:
   - Butterworth (low-pass, default)
   - Kalman (predictive smoothing)
   - Butterworth on speed
   - Gaussian, LOESS, Median


 Filter type and parameters are set in Config.toml under [filtering].
""")


t0 = time.time()
try:
   Pose2Sim.filtering()
   step_times['filtering'] = time.time() - t0
   print(f"n Filtering complete! ({step_times['filtering']:.1f}s)")
except Exception as e:
   print(f" Filtering note: {e}")
   step_times['filtering'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 7/8: MARKER AUGMENTATION")
print("█" * 60)
print("""
 Uses a Stanford-trained LSTM to predict 47 virtual markers from
 the triangulated keypoints. These additional markers can improve
 OpenSim model scaling and inverse kinematics.


 NOTE: Results are NOT always better with augmentation (~50/50).
 The pipeline lets you compare with and without augmentation.
""")


t0 = time.time()
try:
   Pose2Sim.markerAugmentation()
   step_times['markerAugmentation'] = time.time() - t0
   print(f"n Marker augmentation complete! ({step_times['markerAugmentation']:.1f}s)")


   aug_files = list((work_dir / "pose-3d").glob("*augmented*.trc"))
   for af in aug_files:
       print(f"   {af.name} ({af.stat().st_size/1024:.1f} KB)")
except Exception as e:
   print(f" Marker augmentation note: {e}")
   step_times['markerAugmentation'] = time.time() - t0


print("n" + "█" * 60)
print("█  STEP 8/8: OPENSIM KINEMATICS")
print("█" * 60)
print("""
 Performs automatic model scaling and inverse kinematics:
   1. Scale: Adjusts a generic OpenSim model to participant dimensions
      (based on segment lengths from the slowest, most stable frames)
   2. IK: Computes 3D joint angles by fitting the model to markers
  Output:
   - Scaled .osim model file
   - Joint angles .mot file (can be loaded in OpenSim GUI)
""")


t0 = time.time()
if opensim_available:
   try:
       Pose2Sim.kinematics()
       step_times['kinematics'] = time.time() - t0
       print(f"n Kinematics complete! ({step_times['kinematics']:.1f}s)")


       kin_dir = work_dir / "kinematics"
       if kin_dir.exists():
           for f in sorted(kin_dir.glob("*")):
               print(f"   {f.name} ({f.stat().st_size/1024:.1f} KB)")
   except Exception as e:
       print(f" Kinematics error: {e}")
       step_times['kinematics'] = time.time() - t0
else:
   print("   Skipped — OpenSim not installed on this Colab runtime.")
   print("  To run kinematics locally, install via:")
   print("    conda install -c opensim-org opensim")
   step_times['kinematics'] = 0


print("n" + "═" * 60)
print("    PIPELINE TIMING SUMMARY")
print("═" * 60)
total = 0
for step, t in step_times.items():
   bar = "█" * int(t / max(step_times.values(), default=1) * 30)
   print(f"  {step:<22s} {t:6.1f}s  {bar}")
   total += t
print(f"  {'TOTAL':<22s} {total:6.1f}s")
print("═" * 60)

import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
from pathlib import Path
import re


def parse_trc(trc_path):
   """Parse a .trc file and return marker names, frame data, and metadata."""
   with open(trc_path, 'r') as f:
       lines = f.readlines()


   meta_keys = lines[2].strip().split('t')
   meta_vals = lines[3].strip().split('t')
   metadata = dict(zip(meta_keys, meta_vals))


   marker_line = lines[3].strip().split('t')


   header_lines = 0
   for i, line in enumerate(lines):
       if line.strip() and not line.startswith(('PathFileType', 'DataRate',
                                                 'Frame', 't')):
           try:
               float(line.strip().split('t')[0])
               header_lines = i
               break
           except ValueError:
               continue


   raw_markers = lines[3].strip().split('t')
   markers = [m for m in raw_markers if m and m not in ('Frame#', 'Time')]


   marker_names = []
   for m in markers:
       if m and (not marker_names or m != marker_names[-1]):
           marker_names.append(m)


   data_lines = lines[header_lines:]
   data = []
   for line in data_lines:
       vals = line.strip().split('t')
       if len(vals) > 2:
           try:
               row = [float(v) if v else np.nan for v in vals]
               data.append(row)
           except ValueError:
               continue


   data = np.array(data)
   return marker_names, data, metadata




trc_files = sorted((work_dir / "pose-3d").glob("*.trc"))
print(f" Found {len(trc_files)} TRC file(s):")
for f in trc_files:
   print(f"   {f.name}")


trc_file = None
for f in trc_files:
   if 'filt' in f.name.lower() and 'augm' not in f.name.lower():
       trc_file = f
       break
if trc_file is None and trc_files:
   trc_file = trc_files[0]


if trc_file:
   print(f"n Visualizing: {trc_file.name}")
   marker_names, data, metadata = parse_trc(trc_file)
   print(f"   Markers: {len(marker_names)}")
   print(f"   Frames:  {data.shape[0]}")
   print(f"   Marker names: {marker_names[:10]}{'...' if len(marker_names) > 10 else ''}")


   frames = data[:, 0].astype(int) if data.shape[1] > 0 else []
   times = data[:, 1] if data.shape[1] > 1 else []
   coords = data[:, 2:]


   n_markers = len(marker_names)


   mid_frame = len(data) // 2
   fig = plt.figure(figsize=(16, 6))


   ax1 = fig.add_subplot(131, projection='3d')
   xs = coords[mid_frame, 0::3][:n_markers]
   ys = coords[mid_frame, 1::3][:n_markers]
   zs = coords[mid_frame, 2::3][:n_markers]


   ax1.scatter(xs, ys, zs, c='dodgerblue', s=40, alpha=0.8, edgecolors='navy')
   for i, name in enumerate(marker_names[:len(xs)]):
       if i % 3 == 0:
           ax1.text(xs[i], ys[i], zs[i], f' {name}', fontsize=6, alpha=0.7)


   ax1.set_xlabel('X (m)')
   ax1.set_ylabel('Y (m)')
   ax1.set_zlabel('Z (m)')
   ax1.set_title(f'3D Keypoints (Frame {int(frames[mid_frame])})', fontsize=10)


   ax2 = fig.add_subplot(132)
   key_markers = ['RAnkle', 'LAnkle', 'RWrist', 'LWrist', 'Nose']
   colors_map = {'RAnkle': 'red', 'LAnkle': 'blue', 'RWrist': 'orange',
                 'LWrist': 'green', 'Nose': 'purple'}


   for mkr in key_markers:
       if mkr in marker_names:
           idx = marker_names.index(mkr)
           z_col = idx * 3 + 2
           if z_col < coords.shape[1]:
               ax2.plot(times, coords[:, z_col],
                        label=mkr, color=colors_map.get(mkr, 'gray'),
                        linewidth=1.2, alpha=0.8)


   ax2.set_xlabel('Time (s)')
   ax2.set_ylabel('Z position (m)')
   ax2.set_title('Vertical Trajectories', fontsize=10)
   ax2.legend(fontsize=8, loc='best')
   ax2.grid(True, alpha=0.3)


   ax3 = fig.add_subplot(133)
   if len(times) > 1:
       dt = np.diff(times)
       dt[dt == 0] = 1e-6
       for mkr in ['RAnkle', 'RWrist']:
           if mkr in marker_names:
               idx = marker_names.index(mkr)
               x_col, y_col, z_col = idx * 3, idx * 3 + 1, idx * 3 + 2
               if z_col < coords.shape[1]:
                   dx = np.diff(coords[:, x_col])
                   dy = np.diff(coords[:, y_col])
                   dz = np.diff(coords[:, z_col])
                   speed = np.sqrt(dx**2 + dy**2 + dz**2) / dt
                   speed = np.clip(speed, 0, np.nanpercentile(speed, 99))
                   ax3.plot(times[1:], speed, label=mkr,
                            color=colors_map.get(mkr, 'gray'),
                            linewidth=0.8, alpha=0.7)


       ax3.set_xlabel('Time (s)')
       ax3.set_ylabel('Speed (m/s)')
       ax3.set_title('Marker Speeds (Quality Check)', fontsize=10)
       ax3.legend(fontsize=8)
       ax3.grid(True, alpha=0.3)


   plt.tight_layout()
   plt.savefig(work_dir / 'trajectory_analysis.png', dpi=150, bbox_inches='tight')
   plt.show()
   print(" Trajectory plots saved to trajectory_analysis.png")


else:
   print(" No TRC file found to visualize.")

def parse_mot(mot_path):
   """Parse an OpenSim .mot file and return column names and data."""
   with open(mot_path, 'r') as f:
       lines = f.readlines()


   data_start = 0
   for i, line in enumerate(lines):
       if 'endheader' in line.lower():
           data_start = i + 1
           break


   col_line = lines[data_start].strip().split('t')
   if not col_line[0].replace('.', '').replace('-', '').replace('e', '').isdigit():
       col_names = col_line
       data_start += 1
   else:
       col_names = lines[data_start - 2].strip().split('t')


   data = []
   for line in lines[data_start:]:
       vals = line.strip().split('t')
       try:
           row = [float(v) for v in vals]
           data.append(row)
       except ValueError:
           continue


   return col_names, np.array(data)




mot_files = sorted((work_dir / "kinematics").glob("*.mot")) if (work_dir / "kinematics").exists() else []


if mot_files:
   mot_file = mot_files[0]
   print(f" Visualizing joint angles: {mot_file.name}")


   col_names, jdata = parse_mot(mot_file)
   print(f"   Columns: {len(col_names)}")
   print(f"   Frames:  {jdata.shape[0]}")
   print(f"   Joints:  {col_names[:8]}...")


   fig, axes = plt.subplots(2, 3, figsize=(16, 8))
   fig.suptitle('OpenSim Joint Angles from Pose2Sim', fontsize=14, fontweight='bold')


   joints_to_plot = [
       ('hip_flexion_r', 'Right Hip Flexion'),
       ('hip_flexion_l', 'Left Hip Flexion'),
       ('knee_angle_r', 'Right Knee Angle'),
       ('knee_angle_l', 'Left Knee Angle'),
       ('ankle_angle_r', 'Right Ankle Angle'),
       ('ankle_angle_l', 'Left Ankle Angle'),
   ]


   time_col = jdata[:, 0] if 'time' in col_names[0].lower() else np.arange(jdata.shape[0])


   for ax, (joint_key, title) in zip(axes.flat, joints_to_plot):
       col_idx = None
       for i, cn in enumerate(col_names):
           if joint_key.lower() in cn.lower():
               col_idx = i
               break


       if col_idx is not None and col_idx < jdata.shape[1]:
           ax.plot(time_col, jdata[:, col_idx], 'b-', linewidth=1.2)
           ax.set_title(title, fontsize=10)
           ax.set_xlabel('Time (s)')
           ax.set_ylabel('Angle (°)')
           ax.grid(True, alpha=0.3)
       else:
           ax.text(0.5, 0.5, f'{title}n(not found)', ha='center',
                   va='center', transform=ax.transAxes, fontsize=10, color='gray')
           ax.set_title(title, fontsize=10, color='gray')


   plt.tight_layout()
   plt.savefig(work_dir / 'joint_angles.png', dpi=150, bbox_inches='tight')
   plt.show()
   print(" Joint angle plots saved to joint_angles.png")


else:
   print(" No .mot files found (kinematics step was skipped or OpenSim unavailable).")
   print("  Joint angle visualization will be available when running locally with OpenSim.")


print(" DATA QUALITY ANALYSIS")
print("=" * 60)


if trc_file:
   marker_names, data, metadata = parse_trc(trc_file)
   coords = data[:, 2:]


   print("n Per-Marker Data Completeness:")
   print("-" * 45)
   n_frames = data.shape[0]
   for i, mkr in enumerate(marker_names):
       x_col = i * 3
       if x_col < coords.shape[1]:
           valid = np.sum(~np.isnan(coords[:, x_col]))
           pct = 100 * valid / n_frames
           bar = "█" * int(pct / 5) + "░" * (20 - int(pct / 5))
           status = "" if pct > 90 else "" if pct > 50 else ""
           print(f"  {status} {mkr:<20s} {bar} {pct:5.1f}% ({valid}/{n_frames})")


   print("n Trajectory Smoothness (lower = better):")
   print("-" * 45)
   for mkr in marker_names[:10]:
       idx = marker_names.index(mkr)
       cols = [idx * 3, idx * 3 + 1, idx * 3 + 2]
       if cols[2] < coords.shape[1]:
           xyz = coords[:, cols]
           accel = np.diff(xyz, axis=0, n=2)
           jitter = np.nanmean(np.abs(accel))
           bar = "█" * min(int(jitter * 500), 20)
           print(f"  {mkr:<20s} jitter={jitter:.4f}  {bar}")

print("n ADVANCED: PROGRAMMATIC CONFIGURATION")
print("=" * 60)
print("""
You can modify ANY Config.toml parameter programmatically.
This is useful for batch experiments or parameter sweeps.
Below are common modifications for different scenarios.
""")


import toml


config = toml.load(work_dir / "Config.toml")


print("Example 1: Change filter type to Kalman")
example_config_1 = {
   'filtering': {
       'type': 'kalman',
   }
}
print(f"  config['filtering']['type'] = 'kalman'")


print("nExample 2: Tighten triangulation quality thresholds")
example_config_2 = {
   'triangulation': {
       'likelihood_threshold': 0.4,
       'reproj_error_threshold_triangulation': 15,
       'min_cameras_for_triangulation': 3,
   }
}
for k, v in example_config_2['triangulation'].items():
   print(f"  config['triangulation']['{k}'] = {v}")


print("nExample 3: Use lightweight mode for faster processing")
print("  config['pose']['mode'] = 'lightweight'")
print("  config['pose']['mode'] = 'balanced', 'performance', or custom dict")


print("nExample 4: Enable multi-person analysis")
print("  config['project']['multi_person'] = True")
print("  config['project']['participant_height'] = [1.72, 1.65]")
print("  config['project']['participant_mass'] = [70.0, 55.0]")


print("nExample 5: Process only specific frames")
print("  config['project']['frame_range'] = [50, 200]")


print("n To apply changes, modify and save Config.toml, then re-run the pipeline:")
print("""
   import toml
   config = toml.load('Config.toml')
   config['filtering']['type'] = 'kalman'
   with open('Config.toml', 'w') as f:
       toml.dump(config, f)
  
   from Pose2Sim import Pose2Sim
   Pose2Sim.filtering()
""")


print("n ALTERNATIVE: ONE-LINE PIPELINE EXECUTION")
print("=" * 60)
print("""
Instead of calling each step individually, use:


   from Pose2Sim import Pose2Sim
   Pose2Sim.runAll()


Or with selective steps:


   Pose2Sim.runAll(
       do_calibration=True,
       do_poseEstimation=True,
       do_synchronization=True,
       do_personAssociation=True,
       do_triangulation=True,
       do_filtering=True,
       do_markerAugmentation=False,
       do_kinematics=True
   )


This is equivalent to calling each step in sequence, but more concise
for production workflows.
""")


print("n ALL GENERATED OUTPUT FILES")
print("=" * 60)


output_dirs = ['pose', 'pose-3d', 'kinematics']
for dirname in output_dirs:
   d = work_dir / dirname
   if d.exists():
       files = list(d.rglob("*"))
       files = [f for f in files if f.is_file()]
       total_size = sum(f.stat().st_size for f in files) / (1024 * 1024)
       print(f"n {dirname}/ ({len(files)} files, {total_size:.1f} MB total)")
       for f in sorted(files)[:15]:
           rel = f.relative_to(d)
           size = f.stat().st_size / 1024
           print(f"   {''} {rel} ({size:.1f} KB)")
       if len(files) > 15:
           print(f"   ... and {len(files) - 15} more files")


print("n" + "=" * 60)
print(" TO DOWNLOAD RESULTS IN GOOGLE COLAB:")
print("=" * 60)
print("""
 from google.colab import files
 files.download('/content/Pose2Sim_Tutorial/pose-3d/YOUR_FILE.trc')


 !zip -r /content/pose2sim_results.zip /content/Pose2Sim_Tutorial/pose-3d /content/Pose2Sim_Tutorial/kinematics
 files.download('/content/pose2sim_results.zip')
""")


print("""
╔══════════════════════════════════════════════════════════════════════════════╗
║                         TUTORIAL COMPLETE!                               ║
╠══════════════════════════════════════════════════════════════════════════════╣
║                                                                            ║
║  WHAT YOU ACHIEVED:                                                        ║
║   Installed Pose2Sim on Google Colab                                     ║
║   Ran camera calibration (conversion mode)                               ║
║   Performed 2D pose estimation with RTMPose                              ║
║   Synchronized multi-camera views                                        ║
║   Associated persons across cameras                                      ║
║   Triangulated 2D→3D keypoints                                           ║
║   Filtered 3D trajectories                                               ║
║   Augmented markers with Stanford LSTM                                   ║
║   Computed joint angles with OpenSim IK (if available)                   ║
║   Visualized trajectories, joint angles, and data quality                ║
║                                                                            ║
║  NEXT STEPS:                                                               ║
║  • Try with YOUR OWN multi-camera videos                                   ║
║  • Experiment with 'performance' mode for higher accuracy                  ║
║  • Try multi-person mode (Demo_MultiPerson)                                ║
║  • Visualize in OpenSim GUI or Blender (Pose2Sim_Blender add-on)          ║
║  • Use custom DeepLabCut models for animal tracking                        ║
║  • Run batch processing over multiple trials                               ║
║                                                                            ║
║  CITATION:                                                                 ║
║  Pagnon, D., Domalain, M., & Reveret, L. (2022).                         ║
║                                                                            ║
║  GitHub:  https://github.com/perfanalytics/pose2sim                        ║
║  Discord: https://discord.com/invite/4mXUdSFjmt                           ║
║  Docs:    https://perfanalytics.github.io/pose2sim/                        ║
╚══════════════════════════════════════════════════════════════════════════════╝
""")