Map preparation
Preparing a map for automatic detection
Automatic detection traces the lines drawn on your map to find each grave. How the map is captured matters more than anything else you can change — the same plan photographed two different ways can go from a handful of outlines to nearly all of them.
If you read one thing: lay the map flat, shoot it square-on, and upload one section at a time. Folds and angles cause more failures than resolution, file format or lighting combined.
1. Capture
Do
- Flatten the sheet completely. Weigh down the corners if it has been folded or rolled.
- Shoot straight down, camera parallel to the paper, map filling the frame.
- Use a flatbed scanner if you have one. Scans consistently detect better than photographs.
- Even, indirect light — a bright room or open shade.
- Keep the map upright, the way you would read it.
Avoid
- Fold creases. A crease is a dark line across the page, and detection cannot tell it from a drawn boundary.
- Angled or handheld-tilted shots. Perspective bends straight lines and cells stop meeting properly.
- Hard shadows and glare, including your own shadow and window reflections on glossy paper.
- Curled or propped paper — a lifted edge opens gaps in the boundary lines.
Folds are the single most common cause of wrong outlines we see. Where a crease crosses a block, detection tends to produce a lopsided shape that spills into the neighbouring grave — because along that line the map genuinely does look like it has an extra boundary drawn on it.
2. What kind of map works
Detection looks for graves drawn as closed shapes with visible boundary lines — a survey plan, a block layout, a hand-drawn section map. It works best when:
- each grave is enclosed by lines on all sides;
- the lines are unbroken, or nearly so;
- graves are laid out in blocks or rows rather than scattered.
It cannot work from an aerial or satellite photograph — there are no drawn boundaries to trace, and detection will tell you so rather than guess. Maps where the plot dividers are dotted, dashed, or open-ended detect considerably worse, because the shapes never close.
3. Split large maps into sections
Detection now handles a whole sheet far better than it used to: it works through the page in tiles, so outlines on a large, clearly drawn plan are usually found without splitting. Splitting is still worth doing in two cases.
1. More than about 300 graves. Plot codes are read for the first 300 plots on a map only; any beyond that get placeholder codes you would have to fill in by hand. Keep each file under that.
2. A very dense, faint or low-resolution sheet. Smaller sections give the code-reading step larger, clearer numbers, which makes misread codes less likely.
Upload each section under Maps in the plot editor and run detection on one at a time.
Cut on the gaps, never through a grave
This matters more than where exactly you split. Detection finds graves by their enclosed shape, so a grave sliced in half by the edge of your crop is no longer enclosed and is skipped entirely. Cut along the paths and gaps between blocks, and leave a little blank paper around the edge of each section.
A crop taken through the middle of a row of graves returned no graves at all in testing — only the small boxed numbers, which were the only closed shapes left inside the image.
You still get one map at the end
Splitting is for detection only — it does not leave you with a pile of disconnected sections. Once the sections are mapped, Combined view shows them reassembled as one picture, with every plot from every section in its right place, and you can keep editing there.
Sections cut with the Split tool inside the plot editor are positioned automatically. If you cut your map up in another program and uploaded the pieces separately, open Maps and use Place on each one to drag it into position — onto your whole-cemetery image, or onto an empty frame if you never had one.
4. Resolution and file type
| Setting | Recommended | Why |
|---|---|---|
| Format | JPEG or PNG | PDFs work, but only the first page is analysed. |
| Scan resolution | 200–300 dpi | Enough to keep thin boundary lines continuous. |
| Image size | Roughly 2–10 megapixels per section | Below this, thin lines break up. Above it, very large files get a reduced pass to stay within the server's limits. |
| Colour | Leave as captured | Colour, greyscale and black-and-white all work. Do not convert. |
| Orientation | Upright, as you would read it | Phone rotation is corrected automatically; you do not need to fix it. |
Bigger is not automatically better. A 20-megapixel photograph of a whole cemetery detects worse than a 4-megapixel photograph of one block, because what matters is how clearly each individual grave is drawn, not how many pixels the sheet has.
5. Do not "clean up" the image first
This is the most common well-intentioned mistake, and we have measured each of these:
Do not convert to pure black and white
Thresholding an image in an editor destroys the faint and broken lines that detection relies on. Across every map we tested this was severely damaging — on one plan it cut the graves found from 57 to 17, on another from 156 to 9.
Do not erase the plot numbers or names
You do not need to. Detection already removes printed text itself when doing so helps. On most maps, removing the text makes no difference or slightly hurts — the text is not usually what confuses it.
Brightness and contrast: only as a second attempt
Leave the image as captured for the first run. Detection normalises brightness and tries a range of settings on its own, and on most maps the untouched original gave the best result. It is worth trying only if a faint or washed-out scan comes back with a low-confidence warning: on one pale scan, raising contrast turned a low-confidence result into a confident one. Use the Crop / Adjust tool in the plot editor rather than an external editor, so the original is preserved.
6. What to expect
Automatic detection is a starting point, not a finished map. On a good scan of a clearly drawn section it will find most graves. On a folded photograph of a whole cemetery it may find half, with some outlines the wrong shape.
Always review the outlines before saving. Nothing is written to your records until you accept it, and in the plot editor you can drag any corner to correct a shape, or draw missing graves by hand.
7. If the result is poor
| What you see | Most likely cause | What to try |
|---|---|---|
| Small boxes on the plot numbers instead of the graves | The graves are cut off by the edge of the image, so the numbers are the only closed shapes left | Re-crop so whole graves sit inside the image, with blank paper around the edge |
| Lopsided shapes spilling into the next grave | A fold, crease or shadow crossing the block | Flatten the sheet and re-shoot; adjust corners by hand for the few affected |
| Whole rows or blocks missing | Broken boundary lines in that area, or a very large sheet | Split into smaller sections and run each separately |
| Hundreds of tiny fragments | The image is too small or too low-resolution for its detail | Re-scan at higher resolution, or split it into smaller sections |
| A low-confidence warning | Detection judged its own result unreliable | Take it seriously — review carefully, and try a smaller section |
| "This looks like an aerial photo" | No drawn boundaries to trace | Use a drawn plan, or map the graves by hand |
Still stuck? Send us the map and we will tell you what would work best — contact support, or read the FAQ.