Object Tracking¶

This guide covers creating, managing, and analyzing object tracks in VISTA. Tracks represent object trajectories across multiple frames, linking detections or manual observations to form continuous motion paths.

Overview¶

What are Tracks?¶

A Track represents a single object’s trajectory across multiple frames. While detections identify objects at specific points in time, tracks link these observations to form continuous motion paths. Each track contains:

Temporal coordinates: Frame numbers or timestamps
Spatial coordinates: Pixel positions (row/column) or geodetic coordinates (lat/lon/alt)
Visualization properties: Color, marker style, line width, tail length
Motion metadata: Velocity estimates, quality metrics, length
Analysis data: Optional extraction metadata (image chips, signal masks)

Tracks differ from detections in that they:

Represent motion over time rather than static observations
Support gap interpolation and smoothing operations
Track single objects rather than multiple independent observations
Enable trajectory analysis and prediction

Tracks in VISTA¶

VISTA provides two ways to create tracks:

Manual track creation: Click points across frames to manually define trajectories
Automated tracking: Use tracking algorithms to link detections into tracks

The application stores tracks in a hierarchical structure where each Sensor can have multiple Trackers, and each Tracker contains multiple Tracks. This organization allows you to:

Compare results from different tracking algorithms
Maintain tracks from different sensors separately
Export and import tracks in CSV or HDF5 format
Apply visualization and analysis operations to entire trackers

Manual Track Creation¶

Creating Tracks Manually¶

To create a track manually by clicking points across frames:

Open the Data Manager panel (Ctrl+D or View menu)
Select the Tracks tab
Select a Sensor from the dropdown
Click Manual Track button
Navigate through frames and click to add track points: - Left-click to add a point at the clicked location - Re-click a point to remove it
Uncheck the Manual Track tool to complete the track.

Tip

Use keyboard shortcuts for efficient manual tracking:

Use left/right or A/D keys to navigate frames
Click near expected object locations
Use the zoom tool (mouse wheel) for precise placement
Click on a point again to delete it

Loading Tracks from CSV¶

To load tracks from a CSV file:

In the Tracks panel, select a Sensor
Click Load Tracks button
Browse to your CSV file
Tracks are loaded and added to the selected Sensor

See CSV Format for CSV format details.

Tracking Algorithms¶

VISTA provides four tracking algorithms for automatically linking detections into tracks. Access these through Algorithms > Tracking in the main menu.

Simple Tracker¶

The Simple Tracker is a nearest-neighbor tracker that uses frame-to-frame association with velocity prediction. It’s the easiest algorithm to use and requires minimal configuration.

How it Works

Associates detections between consecutive frames based on proximity
Uses linear velocity prediction for position estimates
Creates new tracks for unassociated detections
Deletes tracks after too many consecutive missed detections
Automatically adapts search radius based on detection density

Parameters

Tracker Name: Name for the output tracker
Input Detectors: Select which detectors to use as input
Max Search Radius: Maximum distance (pixels) to search for associations. If not provided, automatically computed from detection statistics
Max Age: Maximum consecutive frames without detection before track deletion. If not provided, automatically computed from frame gaps
Min Track Length: Minimum number of detections required for a valid track (default: 5)

When to Use

Fast-moving objects with consistent motion
Scenes with low to moderate detection density
When you want automatic parameter adaptation
As a baseline for comparison with more sophisticated algorithms

Example Usage

from vista.algorithms.trackers.simple_tracker import run_simple_tracker

config = {
    'tracker_name': 'Simple Tracker',
    'max_search_radius': 20.0,      # Optional: auto-computed if not provided
    'max_age': 5,                    # Optional: auto-computed if not provided
    'min_track_length': 5
}

# Run tracker on detectors
track_data_list = run_simple_tracker(detectors, config)

Kalman Tracker¶

The Kalman Tracker uses a constant velocity Kalman filter for state estimation, providing better prediction in the presence of measurement noise.

How it Works

Maintains state estimate (position + velocity) with uncertainty covariance
Predicts future positions using constant velocity motion model
Associates detections using Mahalanobis distance (accounts for uncertainty)
Updates state estimates when detections are associated
Uses tentative track confirmation based on detection rate

Parameters

Tracker Name: Name for the output tracker
Input Detectors: Select which detectors to use as input
Process Noise: Process noise parameter for motion model (default: 1.0). Higher values allow more velocity variation
Measurement Noise: Measurement noise covariance (default: 1.0). Higher values indicate less trust in detection positions
Gating Distance: Mahalanobis distance threshold for association (default: 5.0). Larger values allow more distant associations
Min Detections: Minimum detections before track confirmation (default: 3)
Delete Threshold: Covariance trace threshold for track deletion (default: 100.0). Tracks with high uncertainty are deleted

When to Use

Noisy detection positions
Objects with relatively constant velocity
Scenes where prediction quality matters
When you need probabilistic state estimation

Example Usage

from vista.algorithms.trackers.kalman_tracker import run_kalman_tracker

config = {
    'tracker_name': 'Kalman Tracker',
    'process_noise': 1.0,
    'measurement_noise': 1.0,
    'gating_distance': 5.0,
    'min_detections': 3,
    'delete_threshold': 100.0
}

track_data_list = run_kalman_tracker(detectors, config)

Network Flow Tracker¶

The Network Flow Tracker formulates tracking as a minimum-cost flow optimization problem, finding globally optimal track assignments rather than greedy frame-to-frame associations.

How it Works

Builds a graph where nodes are detections and edges represent possible associations
Assigns costs to edges based on distance, smoothness, and temporal gaps
Adds entrance/exit costs to control track initiation/termination
Uses Bellman-Ford algorithm to find minimum-cost flow from source to sink
Reconstructs tracks from flow solution

Key Features

Negative link costs incentivize longer tracks over many short tracks
Smoothness penalties encourage constant-velocity motion
Global optimization finds better solutions than greedy association
Gap bridging can link detections across multiple frames

Parameters

Tracker Name: Name for the output tracker
Input Detectors: Select which detectors to use as input
Max Gap: Maximum frame gap to search for associations (default: 5)
Max Distance: Maximum spatial distance for associations (default: 50.0 pixels)
Entrance Cost: Cost for starting a new track (default: 50.0). Higher values prefer longer tracks
Exit Cost: Cost for ending a track (default: 50.0). Higher values prefer longer tracks
Min Track Length: Minimum detections for a valid track (default: 3)

When to Use

Complex scenarios with occlusions or detection gaps
When global optimization is preferred over local greedy decisions
Scenarios where you want to minimize fragmented tracks
When track smoothness is important

Example Usage

from vista.algorithms.trackers.network_flow_tracker import run_network_flow_tracker

config = {
    'tracker_name': 'Network Flow',
    'max_gap': 5,
    'max_distance': 50.0,
    'entrance_cost': 50.0,
    'exit_cost': 50.0,
    'min_track_length': 3
}

track_data_list = run_network_flow_tracker(detectors, config)

Tracklet Tracker¶

The Tracklet Tracker uses a two-stage hierarchical approach optimized for high false alarm scenarios. It first forms high-confidence “tracklets” with strict criteria, then links tracklets using global optimization.

How it Works

Stage 1: Tracklet Formation

Associates detections using strict criteria (small search radius, velocity consistency)
Implements “M out of N” logic (allows small detection gaps)
Filters tracklets by detection rate and quality score
Only retains high-confidence trajectory segments

Stage 2: Tracklet Linking

Extrapolates tracklets forward/backward in time
Scores potential links based on position and velocity consistency
Uses Hungarian algorithm for optimal non-overlapping links
Combines linked tracklets into final tracks

Parameters

Tracker Name: Name for the output tracker
Input Detectors: Select which detectors to use as input
Initial Search Radius: Max distance for tracklet formation (default: 10.0 pixels). Keep small to avoid false associations
Max Velocity Change: Max velocity change for tracklet formation (default: 5.0 pixels/frame). Enforces smooth motion
Min Tracklet Length: Minimum detections for valid tracklet (default: 3)
Max Consecutive Misses: Max consecutive missed detections in Stage 1 (default: 2)
Min Detection Rate: Minimum detection rate (hits/age) for tracklets (default: 0.6)
Max Linking Gap: Maximum frame gap to link tracklets (default: 10)
Linking Search Radius: Max distance for tracklet linking (default: 30.0 pixels)
Smoothness Weight: Weight for smoothness in linking cost (default: 1.0)
Min Track Length: Minimum detections for final track (default: 5)

When to Use

High false alarm scenarios (100:1 false alarm ratio or higher)
Real targets with smooth, consistent motion
When you want to filter out spurious detections early
Scenarios where tracklet-based reasoning is beneficial

Example Usage

from vista.algorithms.trackers.tracklet_tracker import run_tracklet_tracker

config = {
    'tracker_name': 'Tracklet Tracker',
    'initial_search_radius': 10.0,
    'max_velocity_change': 5.0,
    'min_tracklet_length': 3,
    'max_consecutive_misses': 2,
    'min_detection_rate': 0.6,
    'max_linking_gap': 10,
    'linking_search_radius': 30.0,
    'smoothness_weight': 1.0,
    'min_track_length': 5
}

track_data_list = run_tracklet_tracker(detectors, config)

Algorithm Comparison¶

Tracking Algorithm Comparison¶
Algorithm	Best For	Strengths	Weaknesses	Complexity
Simple	Fast objects, low clutter	Fast, automatic parameters	Local associations only	Low
Kalman	Noisy measurements	Probabilistic, smooth predictions	Assumes constant velocity	Medium
Network Flow	Complex scenarios, gaps	Global optimization, gap bridging	Higher computational cost	High
Tracklet	High false alarms	Robust to clutter, two-stage filtering	Many parameters to tune	High

Tracks Panel Overview¶

The Tracks Panel in the Data Manager provides tools for viewing, modifying, and analyzing tracks. The Tracks panel displays all tracks for the selected Sensor in a table with columns:

Visible: Checkbox to show/hide track
Name: Track identifier
Tracker: Parent tracker name
Frames: Number of frames in track
Length: Total trajectory length in pixels
Color: Track color (click to change)
Marker: Marker style for current position
Line Width: Width of trajectory line
Marker Size: Size of position marker
Complete: Show entire track regardless of current frame
Show Line: Whether to draw connecting line
Line Style: Line style (solid, dashed, etc.)
Tail Length: Number of previous frames to show (0 = all)
Labels: Text labels for categorization

Bulk Actions¶

Apply property changes to multiple selected tracks:

Select tracks in the table (Ctrl+Click or Shift+Click for multiple)
Choose a Property from the dropdown (Visibility, Color, Marker, etc.)
Set the Value using the appropriate control
Click Apply to Selected

This is useful for:

Hiding multiple tracks at once
Applying consistent styling across tracks
Batch labeling tracks
Setting tail length for multiple tracks

Track Selection Actions¶

These actions work on one or more selected tracks:

Export Tracks¶

Export selected tracks to CSV file with all coordinates and styling.

Copy to Sensor¶

Copy selected tracks to a different sensor or the same sensor (useful for comparing across sensors).

Note

For real world applications, copying tracks between sensors may require mapping tracks in one sensor to another through some transform operation. The functionality built into VISTA directly copies the tracks’ frames, rows, columns. It is left to users to add any transforms needed for their real-world applications.

Merge Selected¶

To combine multiple tracks into one:

Select two or more tracks in the Tracks panel
Click Merge Selected
A new track is created containing all points from selected tracks
Points are automatically sorted by frame number
Original tracks are preserved

Use cases:

Combining fragmented tracks from the same object
Manually fixing tracking gaps
Consolidating results from multiple tracking runs

Warning

Merging tracks does not remove duplicate frames. If tracks overlap in time, you may need to manually edit the merged result.

Delete Selected¶

Remove selected tracks from the tracker

Label Selected¶

Apply labels to selected tracks for categorization

Plot Track Details¶

Open plotting windows showing track details as a static plot or animated over time.

Track Operations¶

These actions require selecting exactly one track.

Split Track¶

To split a track at a specific frame:

Select a single track in the Tracks panel
Navigate to the frame number where the split should occur
Click Split Track

Use cases:

Correcting tracking errors where two objects were merged
Isolating specific trajectory segments for analysis
Removing problematic sections while preserving good data

Edit Track¶

Interactive track editing allows manual refinement:

Select a single track
Press Edit Track
Edit track points:
- Add point: Navigate to frame, Ctrl+Click at desired location
- Delete point: Click on point that already exists
- Navigate: Use left/right arrow keys and A/D to change frames
Uncheck Edit Track to complete edited track

Tips for editing:

Zoom in for precise positioning
Use keyboard navigation for efficiency
Changes are shown in real-time

Track Extraction¶

Track extraction analyzes imagery around track points to extract signal information and refine positions.

What is Track Extraction?¶

For each track point, extraction:

Extracts a square image chip centered on the track position
Detects signal pixels using CFAR-like thresholding
Computes local noise statistics from background annulus
Optionally updates track coordinates to signal blob centroid

Extraction results are stored in the track’s extraction_metadata and can be:

Viewed as overlays in the viewer
Edited interactively to refine signal masks
Exported for external analysis

Running Track Extraction¶

To extract signal information for a track:

Select a single track in the Tracks panel
Press Extract button
Configure extraction parameters:
Chip Parameters
- Chip Radius: Radius of square chip to extract (pixels)
- Chip diameter = 2 × radius + 1
Background Estimation
- Background Radius: Outer radius for noise calculation (pixels)
- Ignore Radius: Inner guard region to exclude (pixels)
- Annulus Shape: Circular or square background region
Detection Parameters
- Threshold Deviation: Number of standard deviations above mean for signal detection
- Search Radius: Only keep signal blobs within central search region (optional)
Centroid Update
- Update Centroids: If checked, refine track positions using signal blob centroids
- Max Centroid Shift: Maximum allowed position update (pixels)
Press Run
Extraction metadata is stored in the track

Viewing Extraction Results¶

To view extracted signal pixels:

Select a track with extraction metadata
Press View Extraction
Navigate through frames to see signal pixel overlay
Signal pixels are highlighted in the viewer
Press View Extraction again to hide overlay

Editing Extraction Results¶

To manually refine signal masks:

Select an extracted track
Press Edit Extraction
Paint or erase signal pixels:
- Paint: Left-click and drag to mark pixels as signal
- Erase: Right-click and drag to remove signal pixels
- Adjust brush: Use mouse wheel to change brush size
- Navigate: Arrow keys or Page Up/Down to change frames
Press Save Changes to update extraction metadata
Press Cancel or press Escape to discard changes

Use cases:

Correcting false positives in signal detection
Adding missed signal pixels
Refining masks before exporting for machine learning

Track Uncertainty Visualization¶

Tracks can optionally include positional uncertainty information, displayed as ellipses around track points.

What is Track Uncertainty?¶

Track uncertainty represents the estimated position error for each track point, typically arising from:

Detection noise and measurement error
Tracking filter state covariance (e.g., Kalman filter)
Interpolation or extrapolation uncertainty
Sensor geolocation accuracy

Covariance Matrix Representation

VISTA stores uncertainty as a 2D position covariance matrix with three elements:

\[\begin{split}\Sigma = \begin{bmatrix} C_{00} & C_{01} \\ C_{01} & C_{11} \end{bmatrix}\end{split}\]

Where: - C₀₀ (Covariance 00): Row variance (pixels²) - C₀₁ (Covariance 01): Row-column covariance (pixels²) - C₁₁ (Covariance 11): Column variance (pixels²)

This format directly matches the output from most tracking algorithms (Kalman filters, particle filters) and unambiguously represents any 2D Gaussian uncertainty distribution.

Visualization

Uncertainty is visualized as an ellipse at each track point, where:

Ellipse axes are derived from the covariance matrix eigenvalues
Ellipse orientation shows the direction of maximum uncertainty
Ellipse color matches the track color
Line style is configurable (default: dashed)

Enabling Uncertainty Display¶

To show uncertainty ellipses for a track:

Ensure track has uncertainty data (loaded from CSV or generated by tracking algorithm)
In the Tracks panel, check the Show Uncertainty checkbox for that track
Ellipses will appear at each visible track point

Note

The “Show Uncertainty” checkbox is only enabled for tracks that have uncertainty data.

Configuring Uncertainty Display¶

Customize uncertainty ellipse appearance in Settings > Track Visualization:

Line Style: Solid, Dashed (default), Dotted, Dash-Dot, Dash-Dot-Dot
Line Width: 1-10 pixels (default: 1)
Scale Factor: 0.1-10.0 (default: 1.0) - multiplier for ellipse size

The scale factor allows you to visualize different confidence levels:

Scale = 1.0: 1-sigma ellipse (~68% confidence)
Scale = 2.0: 2-sigma ellipse (~95% confidence)
Scale = 3.0: 3-sigma ellipse (~99.7% confidence)

Exporting and Importing Tracks¶

CSV Format¶

Tracks can be exported to and imported from CSV files with the following format:

Required Columns

One temporal coordinate (choose one):

Frames: Frame numbers (preferred)
Times: Timestamps in ISO format (requires sensor with imagery times)

One spatial coordinate system (choose one):

Rows and Columns: Pixel coordinates (preferred)
Latitude (deg), Longitude (deg), Altitude (km): Geodetic coordinates (requires sensor with geolocation)

Optional Columns

Track: Track name (required for multi-track files)
Tracker: Tracker name
Color: PyQtGraph color string (e.g., ‘r’, ‘g’, ‘b’, ‘#FF0000’)
Marker: Marker style (‘o’, ‘s’, ‘t’, ‘d’, ‘+’, ‘x’, ‘star’)
Line Width: Integer line width
Marker Size: Integer marker size
Visible: Boolean visibility
Complete: Boolean to show entire track
Show Line: Boolean to draw connecting line
Line Style: Line style string (‘SolidLine’, ‘DashLine’, ‘DotLine’, ‘DashDotLine’, ‘DashDotDotLine’)
Tail Length: Integer (0 = show all history)
Labels: Comma-separated label strings
Covariance 00: Position covariance matrix element (row variance, in pixels²)
Covariance 01: Position covariance matrix element (row-column covariance, in pixels²)
Covariance 11: Position covariance matrix element (column variance, in pixels²)

Example CSV (Pixel Coordinates)

Track,Frames,Rows,Columns,Color,Marker,Visible
Object_1,100,250.5,300.2,r,o,True
Object_1,101,252.3,305.8,r,o,True
Object_1,102,254.1,311.4,r,o,True
Object_2,100,180.2,420.5,g,s,True
Object_2,101,182.5,422.1,g,s,True

Example CSV (Geodetic Coordinates)

Track,Frames,Latitude (deg),Longitude (deg),Altitude (km),Color
Satellite_1,100,38.5,77.0,400.0,y
Satellite_1,101,38.6,77.1,401.0,y
Satellite_1,102,38.7,77.2,402.0,y

Example CSV (Time-based)

Track,Times,Rows,Columns,Labels
Target_A,2024-01-15T10:30:00.000,250.5,300.2,"aircraft,commercial"
Target_A,2024-01-15T10:30:01.000,252.3,305.8,"aircraft,commercial"
Target_A,2024-01-15T10:30:02.000,254.1,311.4,"aircraft,commercial"

Example CSV (With Uncertainty)

Track,Frames,Rows,Columns,Covariance 00,Covariance 01,Covariance 11,Color
Track_1,100,250.5,300.2,4.0,0.5,2.25,r
Track_1,101,252.3,305.8,4.2,0.6,2.5,r
Track_1,102,254.1,311.4,4.5,0.7,2.8,r

Loading from CSV¶

Via GUI:

Open Data Manager > Tracks tab
Select a Sensor
Press Load Tracks
Browse to CSV file
Tracks are loaded and added to the sensor

Via Python:

import pandas as pd
from vista.tracks.track import Track

# Load CSV
df = pd.read_csv('tracks.csv')

# Create tracks from DataFrame
for track_name, track_df in df.groupby('Track'):
    track = Track.from_dataframe(track_df, sensor, name=track_name)
    # Add to sensor or tracker

Exporting to CSV¶

Via GUI:

Select tracks in Tracks panel
Press Export Tracks
Choose save location
Tracks are exported with all columns

Via Python:

# Export single track
track.to_csv('my_track.csv')

# Export tracker with multiple tracks
tracker.to_csv('all_tracks.csv')

Common Workflows¶

Workflow 1: Manual Tracking¶

Create tracks by clicking through frames:

Load imagery for a sensor
Open Data Manager > Tracks tab
Press Manual Track
Navigate through frames, clicking object positions
Press Enter to finish
Repeat for additional objects
Export tracks to CSV when complete

Workflow 2: Automated Tracking¶

Use detection algorithms to create tracks automatically:

Run detection algorithm (e.g., Simple Threshold or CFAR)
Review detections in Data Manager > Detections tab
Go to Algorithms > Tracking menu
Select tracking algorithm (e.g., Simple Tracker)
Configure parameters and select input detectors
Run algorithm
Review tracks in Tracks tab
Refine using split, merge, or edit operations

Workflow 4: Comparative Analysis¶

Compare multiple tracking algorithms:

Run detection algorithm on your data
Run multiple tracking algorithms with different parameters:
- Simple Tracker with default settings
- Kalman Tracker for smooth motion
- Network Flow for global optimization
- Tracklet Tracker for high clutter
Each produces a separate Tracker in the Tracks panel
Use Plot Track Details to compare trajectories
Use Copy to Sensor to compare across sensors
Export best results to CSV

Tips and Best Practices¶

Track Creation Tips¶

Use appropriate tracking algorithm for your scenario:
- Fast objects with low clutter → Simple Tracker
- Noisy detections → Kalman Tracker
- Complex scenarios with gaps → Network Flow
- High false alarms → Tracklet Tracker
Start with conservative parameters:
- Smaller search radius reduces false associations
- Higher minimum track length filters spurious tracks
- Let automatic parameter estimation guide you
Review detection quality first:
- Poor detections → poor tracks
- Filter detections by confidence before tracking
- Adjust detection thresholds if needed

Track Management Tips¶

Use labels for organization:
- Label tracks by object type (aircraft, vehicle, person)
- Label by behavior (inbound, stationary, turning)
- Label by quality (verified, suspicious)
Use tail length for visualization:
- Set tail length = 10-20 for recent history
- Set tail length = 0 to see full trajectory
- Use “Complete” mode to show track regardless of frame
Adjust styling for clarity:
- Use distinct colors for different objects
- Increase line width for long-range objects
- Use dashed lines for predicted/interpolated sections

Analysis Tips¶

Interpolate before smoothing:
- Fill gaps with interpolation first
- Then apply Savitzky-Golay smoothing
- This produces better results than smoothing sparse tracks
Choose appropriate smoothing parameters:
- Larger radius = more smoothing
- Cubic polynomials preserve features better than linear
- Window length should be < 10% of track length
Use extraction for quality assessment:
- Extract chips to verify object presence
- Check noise_std values for track reliability
- Use signal masks to identify tracking errors

Performance Tips¶

Limit track visualization:
- Hide tracks you’re not currently analyzing
- Use tail length to reduce rendering overhead
- Toggle off “Show Line” for many tracks
Use appropriate tracking algorithm complexity:
- Simple Tracker is fastest
- Network Flow is slower but more accurate
- Consider computational cost for large datasets
Filter detections before tracking:
- Remove low-confidence detections
- Apply spatial or temporal filtering
- This improves both speed and accuracy

Troubleshooting¶

No tracks created by algorithm

Check that detectors have sufficient detections
Reduce minimum track length parameter
Increase max search radius or max gap
Verify detection density is adequate

Fragmented tracks

Increase max gap or max age parameters
Use Network Flow tracker for better gap bridging
Consider interpolation to fill gaps post-tracking
Manually merge related track fragments

Track switches (two objects swap identities)

Reduce max search radius to prevent ambiguous associations
Use Kalman or Tracklet tracker for better discrimination
Increase process noise in Kalman tracker
Manually split incorrectly merged tracks

Tracks drift from objects over time

Run extraction to identify tracking errors
Use Edit Track to manually correct positions
Try smoothing with smaller window radius
Check if detections are properly centered on objects

Object Tracking¶

Overview¶

What are Tracks?¶

Tracks in VISTA¶

Manual Track Creation¶

Creating Tracks Manually¶

Loading Tracks from CSV¶

Tracking Algorithms¶

Simple Tracker¶

Kalman Tracker¶

Network Flow Tracker¶

Tracklet Tracker¶

Algorithm Comparison¶

Tracks Panel Overview¶

Bulk Actions¶

Track Selection Actions¶

Export Tracks¶

Copy to Sensor¶

Merge Selected¶

Delete Selected¶

Label Selected¶

Plot Track Details¶

Track Operations¶

Split Track¶

Edit Track¶

Track Extraction¶

What is Track Extraction?¶

Running Track Extraction¶

Viewing Extraction Results¶

Editing Extraction Results¶

Track Uncertainty Visualization¶

What is Track Uncertainty?¶

Enabling Uncertainty Display¶

Configuring Uncertainty Display¶

Track Analysis and Refinement¶

Interpolation¶

Smoothing (Savitzky-Golay)¶

Exporting and Importing Tracks¶

CSV Format¶

Loading from CSV¶

Exporting to CSV¶

Common Workflows¶

Workflow 1: Manual Tracking¶

Workflow 2: Automated Tracking¶

Workflow 3: Track Refinement¶

Workflow 4: Comparative Analysis¶

Tips and Best Practices¶

Track Creation Tips¶

Track Management Tips¶

Analysis Tips¶

Performance Tips¶

Troubleshooting¶

See Also¶