I will present an unsupervised learning-based anomaly detection technique for generic and model-independent new physics searches at the LHC. New physics signals are considered anomalies causing deviations in data with respect to expected background. I will discuss a number of auto-encoders (AEs) trained using raw detector images, which are large and highly sparse in nature—a computer-vision problem. We used these AEs to search for manifestations of a dark version of strong force—producing high-multiplicity dark hadrons (dark showers) in proton collisions—without leveraging any physics-based pre-processing or assumption on the signals. Our proposed AE has a dual-encoder design, which is general, and can learn an auxiliary yet compact latent space through spatial conditioning, showing a neat improvement over competitive physics-based baselines and related approaches, therefore reducing the gap with fully supervised models. It is the first time an unsupervised model has shown to exhibit excellent discrimination against multiple dark shower models, illustrating the suitability of this method as an accurate, fast, model-independent algorithm to deploy, e.g., in the real-time event triggering systems of LHC experiments such as ATLAS and CMS.