Abstract: Systems and methods for an improved bounding box tool. The system can have processors configured to display a user interface on a display of a device, the user interface displaying image data. The processor can activate a virtual tool to define a bounding box. The processor can receive a first input data point at a first location relative to the image data. This can be defined by actuation of the input device. The processor can receive a movement input in a direction relative to the first location. The movement input can be defined by movement commands from the input device. The processor can receive a second input data point at a second location relative to the image data. The second input data point can be triggered by a second actuation of the input device. The processor can display, at the user interface, a graphical object representing the bounding box as an overlay of the image data. The bounding box can have corners defined by the first location and the second location.

Abstract: Systems and methods for an improved bounding box tool. The system can have processors configured to display a user interface on a display of a device, the user interface displaying image data. The processor can activate a virtual tool to define a bounding box. The processor can receive a first input data point at a first location relative to the image data. This can be defined by actuation of the input device. The processor can receive a movement input in a direction relative to the first location. The movement input can be defined by movement commands from the input device. The processor can receive a second input data point at a second location relative to the image data. The second input data point can be triggered by a second actuation of the input device. The processor can display, at the user interface, a graphical object representing the bounding box as an overlay of the image data. The bounding box can have corners defined by the first location and the second location.

Abstract: System and methods for associating unlabeled images with other, labeled images of the same locations. A high-level signature of each image is generated, representing high-level structural features of each image. A signature of an unlabeled image is then compared to a signature of a labeled image. If those signatures match within a margin of tolerance, the images are interpreted as representing the same location. One or more labels from the labeled image can then be automatically applied to the unlabeled image. In one embodiment, the images are frames from separate video sequences. In this embodiment, entire unlabeled video sequences can be labeled based on a labeled video sequence covering the same geographic area. In some implementations, the high-level signatures are generated by rule-based signature-generation modules. In other implementations, the signature-generation module can be a neural network, such as a convolutional neural network.