The lack of annotated semantic segmentation datasets for electrical substations in the literature poses a significant problem for machine learning tasks; before training a model, a dataset is needed. This paper presents a new dataset of electric substations with 1660 images annotated with 15 classes, including insulators, disconnect switches, transformers and other equipment commonly found in substation environments. The images were captured using a combination of human, fixed and AGV-mounted cameras at different times of the day, providing a diverse set of training and testing data for algorithm development. In total, 50,705 annotations were created by a team of experienced annotators, using a standardized process to ensure accuracy across the dataset. The resulting dataset provides a valuable resource for researchers and practitioners working in the fields of substation automation, substation monitoring and computer vision. Its availability has the potential to advance the state of the art in this important area.

Energy losses in transmission lines can be aggravated by contact deterioration and corrosion, posing significant challenges to the reliability of energy distribution. Those conditions can be quickly identified using thermography, a popular non-invasive inspection method mandatory in many countries, with a minimum frequency determined by legislation. While recent advancements have focused on automating this process using artificial intelligence, working directly on the infrared domain remains a challenging task; not only assembling an infrared dataset requires specialists to properly annotate the images, making the task harder and more expensive, mainstream object detectors, made for the visible light domain, require modifications to work correctly on the infrared domain. Our proposed method allows using an object detector with results comparable to other works that performed detection directly on the infrared domain while requiring no significant changes to the model architecture. By employing stereo cameras, capable of collecting both the visible light spectrum and infrared, we can perform the detection on the visible light images and then transfer the box coordinates to the paired infrared image after rectifying the image pair to match.