AR Toolbox

Scan Workflow: From Capture to Review to Save

The scan workflow is the central loop of AR Toolbox. It is the process you will repeat dozens or hundreds of times as you build and maintain your tool inventory. Understanding each step in this workflow, and knowing where you can customize or accelerate it, turns scanning from a task into a habit that takes seconds. This post walks through every stage of the process from the moment you point your camera at a set of tools to the moment those tools are saved in your inventory database.

Aim, Detect, and Capture

The workflow begins on the scan screen, where your phone's camera feed fills the display. Position your phone so the tools you want to scan are clearly visible in the frame. The detection model runs continuously on the live feed, and you will see AR overlays appear in real time as tools are recognized. This live preview lets you adjust your position before committing to a scan. If you notice tools at the edge of the frame are not being detected, shift your phone slightly to bring them fully into view.

When you are satisfied with what the camera sees, tap the capture button or say "scan" using a voice command. This freezes the current frame and runs a final high-confidence detection pass. The reason for this two-stage approach, live preview followed by committed capture, is that the live preview prioritizes speed and may use a slightly lower resolution, while the capture pass processes the full-resolution frame for maximum accuracy. The result is a definitive set of detections with the highest confidence scores the model can produce.

After capture, the screen transitions to show the frozen frame with all detected tools highlighted by bounding boxes. A count badge in the corner tells you how many tools were found. This is your first opportunity to visually verify that the results look reasonable before moving to the detailed review step.

Review, Edit, and Confirm

Tapping the review button opens a list view of every detected tool. Each entry shows the tool name, category color indicator, and confidence score. The list is sorted by confidence, with the most certain detections at the top and the least certain at the bottom. This ordering naturally draws your attention to the items most likely to need correction.

For each entry, you have several options. If the detection is correct, you do nothing and move on. If the tool name is wrong, tap it to open the correction picker, where you can search all 130+ tool types and select the right one. If a detection is a false positive, meaning the model identified something as a tool that is not actually a tool, swipe to dismiss it. If the model missed a tool entirely, tap the add button to manually insert a tool from the catalog.

The review screen also lets you adjust quantities. If the model detected two adjustable wrenches but there are actually three in the frame and one was obscured, you can manually increase the count. Quantity adjustments are common for multiples of the same tool type, especially small items like screwdriver bits or socket adapters that can overlap in a toolbox.

Once you are satisfied with the review, select the target container from the container picker at the top of the screen. If the correct container is already active from a previous scan, this step is automatic. If you need to switch containers or create a new one, the picker provides quick access without leaving the review screen.

Experienced users often skip the detailed review for routine scans of familiar containers. When you scan your everyday toolbox and the count matches what you expect, tapping save without reviewing every line item is perfectly reasonable. The review step is there when you need it, not to slow you down when you don't.

Save and Batch Scanning

Tapping save commits every reviewed item to the inventory database under the selected container. The save operation records each tool with its type, category, quantity, container assignment, and a timestamp. If any of the saved tools already exist in a different container, the system logs a movement event. If a tool already exists in the same container, the quantity is updated to reflect the latest scan. The save operation is nearly instantaneous because everything is written to the local database on your device.

After saving, the app returns to the scan screen ready for the next capture. This is where batch scanning becomes powerful. If you are inventorying a large collection, like all the compartments of a fully loaded service truck, you simply move through them one by one. Scan a compartment, review quickly, save, switch to the next container either by tapping or saying "next container," and repeat. The workflow is designed to be rhythmic and fast, minimizing the time between physical actions, opening a compartment, and digital actions, capturing its contents.

For very large surfaces that cannot fit in a single frame, you can perform multiple captures into the same container. The app merges results intelligently, recognizing when a second scan of the same container surfaces tools that were already saved and avoiding duplicates. New tools found in the second scan are added, and the overall inventory for that container reflects the combined results of all scans. This merge behavior means you do not need to worry about capturing everything in one perfect shot. Scan in sections, and the app handles the aggregation.

The total time for a complete scan-review-save cycle depends on how many tools are in the frame and how much review is needed, but for a typical toolbox drawer with 10 to 15 items, the entire process takes well under a minute. For routine checks of a familiar container where the contents have not changed much, the whole cycle can be completed in 10 to 15 seconds.

What's Next

The scan workflow feeds directly into the reporting and history system, which gives you powerful tools for auditing your inventory, tracking changes over time, and sharing data with your team. In the next post, we will explore reports, scan history timelines, and export capabilities that turn your scan data into actionable business information.

Back to Blog