Deep learning model to track objects in motion imagery. A brief summary of the item is not available. Add a brief summary about the item.
Deep learning package
from
Item created: Mar 16, 2021 Item updated: Jan 9, 2025 Number of downloads: 2,812
Description
Manually digitizing the track of an object can be a slow process. This model automates the object tracking process significantly, and hence speeds up motion imagery analysis workflows. It can be used with the Motion Imagery Toolset found in the Image Analyst extension to track objects. The detailed workflow and description of the object tracking capability in ArcGIS Pro can be found here.
This model can be used for applications such as object follower and surveillance of stationary objects. It does not perform very well in case there are sudden camera shakes or abrupt scale changes.
Using the model
Follow the guide to use the model. The model can be used with the Motion Imagery tools in ArcGIS Pro 2.8 and onwards. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS.
Fine-tuning the model
This model cannot be fine-tuned using ArcGIS tools.
Input
Object to track marked as a bounding box in 8-bit, 3-band high resolution full motion video / motion imagery. Recommended object size is greater than 15x15 (in pixels).
Output
Bounding box depicting object location in successive frames.
Applicable geographies
This model is expected to work well in all regions globally for any generic-type of objects of interest. However, results can vary for motion imagery that are statistically dissimilar to the training data.
Model architecture
This model uses the SiamMask model architecture implemented in ArcGIS API for Python.
Accuracy metrics
The model has an average precision score of 0.853.
Training data
The model was trained using image sequences from the DAVIS dataset licensed under CC BY 4.0 license, and further fine-tuned on aerial motion imagery.
Sample results
Here are a few results from the model.
An in-depth description of the item is not available.
Layers
Tools
Tables
Basemap
Project Contents:
Solution Contents
Contents
Layers
Screenshots
Terms of Use
No special restrictions or limitations on using the item's content have been provided.
Details
Dashboard views: Desktop
Source:
Creating data in:
Published as:
Other Views:
Dependent items in the recycle bin
Applicable: 2d
Size: 78.584 MB
ID: fbf7d003fdfd4605af56b281ab60be17
Image Count: 0
Image Properties
Layer Drawing
Using tiles from a cache
Dynamically from data
Share
Owner
Folder
Categories
This item has not been categorized.
Credits (Attribution)
No acknowledgements.Esri
Comments (1)
Hello, I am trying to load this model using arcgis.learn like so: model = SiamMask.from_model("./ObjectTracking.dlpk"), however, an error pops out: "Exception: Please check model file path". From what I can see, it seems that code from arcgis.learn is loading this model as a single file, but after extracting dlpk I can see this is rather a multi-model pipeline. Of course, there might be other aspects causing the issue. Were there any SiamMask architecture code changes?