Computer vision (CV) has made a huge impact in many scientific disciplines, and the continuous development of new methods will allow researchers to bridge their work across increasingly diverse use cases and domain areas. At the Alan Turing Institute we have developed Scivision – an open-source software package, a catalogue of data and models, and a community of CV experts and users. This platform enables discovery and reuse of CV models and datasets, allowing researchers to explore new techniques for data analysis as well as facilitating discovery of sample datasets for CV algorithm development. One such example is Graph Representation Analysis for Connected Embeddings (GRACE), a Python package for identification of structural motifs in complex imaging data. GRACE sources from large images populated with low-fidelity object detections to build a graph representation of the entire image. Object detection employs a human-in-the-loop approach, allowing the user to annotate motifs of interest, thus making the tool agnostic to object type. GRACE reduces the search space from millions of pixels to hundreds of nodes, which allows for fast and efficient implementation and potential tool customisation. This method can also be repurposed to search for different motifs of interest within the same dataset. This end-to-end approach could be extended to other types of imaging data where object segmentation and detection remain challenging.