Skip to main content
. 2022 Sep 24;19(19):12095. doi: 10.3390/ijerph191912095

Figure 1.

Figure 1

Computer vision model. Each sample is a single image accompanied by labels corresponding to each neighborhood feature (e.g., crosswalk). The feature extractor is VGG-19 or ResNet18 (depending on the feature analyzed) and is pretrained with ImageNet data. Each feature classifier is a single fully connected layer and the losses are cross entropy. The final loss for optimization is a summation of losses.