Retail Intelligence
Visual data for smarter shelves and stores.
Retail computer vision starts with product detection, shelf compliance, and customer flow analysis. cvPal gives retail teams the tools to build, clean, and version the datasets that power these models — without writing data scripts.
Build product detection datasets
Import shelf images from local cameras or pull existing retail datasets from Hugging Face and Kaggle. Annotate with bounding boxes for SKU detection, segmentation masks for planogram compliance, or classification labels for product categorization.
Standardize across store locations
Merge datasets from different stores with varying camera angles and lighting. Automatic class reindexing handles overlapping product labels. Rename 'cola_can' to 'beverage_carbonated' globally, or split 'snacks' into granular subcategories.
Augment for in-store variability
Store lighting changes throughout the day. Augment with brightness, contrast, and rotation variations — all annotation-aware, so bounding boxes and masks transform correctly. Build models that work at 8am and 8pm.
Export for edge deployment
Export as YOLO for in-store edge devices, COCO for cloud inference, or HuggingFace format for model sharing. Every export is tracked in your library with version, format, and class filter metadata.