DAVE is a low-shot object counting method that combines density-based and detection-based approaches using a detect-and-verify paradigm. It first generates a high-recall detection set and then verifies the detections to identify and remove outliers, improving both recall and precision. DAVE outperforms existing methods in total count accuracy, detection quality, and zero-shot and text-prompt-based counting. It achieves a 20% reduction in MAE compared to density-based counters and a 20% improvement in detection metrics over detection-based counters. DAVE also sets a new state-of-the-art in zero-shot and text-prompt-based counting. The method is evaluated on multiple benchmarks, including FSC147 and FSCD147, showing superior performance in both density-based and detection-based counting. DAVE's architecture is effective in handling multiple object classes and is the first zero-shot detection-based counter. The method is also adapted for text-prompt-based counting, demonstrating strong performance in various counting scenarios. The results show that DAVE significantly improves the accuracy of object counting in low-shot settings, particularly in challenging scenarios with high object density or multiple object classes.DAVE is a low-shot object counting method that combines density-based and detection-based approaches using a detect-and-verify paradigm. It first generates a high-recall detection set and then verifies the detections to identify and remove outliers, improving both recall and precision. DAVE outperforms existing methods in total count accuracy, detection quality, and zero-shot and text-prompt-based counting. It achieves a 20% reduction in MAE compared to density-based counters and a 20% improvement in detection metrics over detection-based counters. DAVE also sets a new state-of-the-art in zero-shot and text-prompt-based counting. The method is evaluated on multiple benchmarks, including FSC147 and FSCD147, showing superior performance in both density-based and detection-based counting. DAVE's architecture is effective in handling multiple object classes and is the first zero-shot detection-based counter. The method is also adapted for text-prompt-based counting, demonstrating strong performance in various counting scenarios. The results show that DAVE significantly improves the accuracy of object counting in low-shot settings, particularly in challenging scenarios with high object density or multiple object classes.