Crowd counting from an image is a challenging task due to occlusion, low quality, and scale variation of objects. With the development of deep learning techniques, various crowd counting methods have been proposed in response to this challenge. This model uses a method that solves the crowd counting problem.
To complete this workflow, the following are the license requirements:
- ArcGIS Desktop—ArcGIS Image Analyst extension for ArcGIS Pro
- ArcGIS Enterprise—ArcGIS Image Server
- ArcGIS Online—ArcGIS Image for ArcGIS Online
This model has the following characteristics:
- Input—JPG or JPEG image. The recommended image resolution for optimal results is less than 2000x2000.
- Output—Feature layer with the number of classes as the count of people in the input.
- Compute—This workflow is compute intensive and a GPU with compute capability of 6.0 or higher is recommended.
- Architecture—This model uses Distribution Matching for Crowd Counting (DM-Count).
- Accuracy metrics—Average PSNR and SSIM over the whole QNRF test set are 40.65 and 0.55, respectively.
- Training dataset—This model is trained using the UCF-QNRF dataset.
Access and download the model
Download the Crowd Counting pretrained model from ArcGIS Living Atlas of the World. Alternatively, access the model directly from ArcGIS Pro, or consume it in ArcGIS Image for ArcGIS Online
To download the model, complete the following steps:
The following are the release notes: