Image dataset for AI-based key point detection of cables and wires at electronic connectors

The image dataset enables the training of a neural network to recognise cables and wires at electronic connectors in electronic devices for automated cable separation. It was developed as part of the Desire4Electronics (https://www.ipa.fraunhofer.de/de/referenzprojekte/Desire4Electronics.html) project and is used to recognise cables and wires at plugs or electronic connestors, so that they can be automatically separated/cut by an industrial robot. The dataset comprises 140 images for training and 35 for validation, each with multiple labelled cables and wires at electronic connectors. The key point labels are in YOLO Pose Format with three keypoints provided for each plug (px1, py1, px2, py2, px3, py3). There is just one class index [0], similar for all connectors: <class-index> <x> <y> <width> <height> <px1> <py1> 2 <px2> <py2> 2 ... <pxn> <pyn> 2