However, an image classifier will look at the entire image and is meant to detect one class. A threshold was made for the amount the color could change, and a threshold for the number of pixels that could change color, before it was considered motion.Īs I showed in another project, Convolutional Neural Networks can be used to classify objects in an image. This was done by counting the number of pixels that had their colors change from the previous image. To save storage space and lower computing time, images were only kept if movement was detected. The webcam was set to constantly take images. By tracking whether boxes enter or exit from the left or right, it is possible to tell which truck is being loaded or unloaded. I set up a webcam between two truck ports so I could view boxes coming in and leaving the warehouse. However, some shipments have multiple types of boxes stacked together, so I am also testing the feasibility of differentiating between boxes in a shipment. I am not counting the number of boxes, but rather detecting what boxes are in a shipment.
This was largely a demonstration of what can be tracked, and I started out by detecting and tracking two types of boxes. With this expansion comes a need for automation, so I took on the challenge of tracking what boxes are in shipments that are arriving and departing from their warehouse.
They were founded in 2013 in Chicago, IL, and have already expanded to selling 2.5 million meals per month to people in 48 states. Home Chef is a company that delivers pre-packaged ingredients for meals.
0 kommentar(er)
