Seed counting is finding the exact number of seeds in a sample, packet or lot. It is done by hand, by weight using an average seeds-per-gram figure, or with an automatic seed counter that counts each seed as it passes a sensor or camera. Hand counting is slow, counting by weight is fast but only as accurate as the weight assumption, and an automatic seed counter gives an exact, repeatable count at production speed.
This article is part of our automated counting guide, which covers counting technologies across industries. Here we look at seed specifically: why exact counts matter, how each counting method works, where accuracy is lost, and how to choose an automatic seed counter.
Why exact seed counts matter
- Packeting and labeling. A packet labeled 500 seeds should hold 500 seeds. Short packets bring complaints; over-filled packets give away expensive seed, especially hybrids and pelleted seed.
- Breeding and field trials. Every plot needs the same number of seeds, or yield differences between varieties are partly caused by seeding rate.
- Germination and quality testing. Germination tests are commonly run on 400 seeds in four replicates of 100, so every test starts from an exact count.
- Planting rates and seed weight. Thousand-seed weight and seeds per gram turn a target plant population into kilograms of seed per hectare or pounds per acre.
Seed counting methods compared
| Method | How it works | Strengths | Limitations |
|---|---|---|---|
| Hand counting | Seeds are counted by eye, often with counting boards, trays or vacuum heads that pick up a fixed number | No equipment cost, fine for very small samples | Slow, tiring, and error rates rise over a shift |
| Counting by weight | A sample is weighed and the count is calculated from an average weight per seed | Very fast, uses an existing scale | Assumes every seed weighs the same, which varies between varieties, lots and seed sizes |
| Vibratory or photoelectric counter | Seeds are fed single file past a light beam that counts interruptions | Faster than hand counting, suited to uniform seed | Touching or overlapping seeds and debris can be miscounted; no size data |
| Vision-based automatic seed counter | Seeds fall past a camera that images each seed and counts it in software | Exact counts at high speed, separates overlapping seed, measures each seed | Higher purchase cost than a scale or simple counter |
Why counting by weight goes wrong
Weight counting is attractive because a scale is quick and cheap. The problem is that seed weight varies, so a weight that matches the target count for one lot is wrong for the next.
Worked example: filling a 500-seed packet by weight
The setup assumes an average of 3.0 mg per seed, so the packet target is 500 × 3.0 mg = 1.5 g.
The next lot has larger seed averaging 3.3 mg, 10% heavier. A 1.5 g fill now holds 1,500 mg ÷ 3.3 mg ≈ 455 seeds, about 45 short of the label (9% under).
A lot of smaller seed averaging 2.7 mg gives 1,500 ÷ 2.7 ≈ 556 seeds per packet, about 56 extra (11% given away).
Re-checking the seed weight for every lot reduces the error, but only an actual count makes every packet exact.
The same trade-off appears in manufacturing, as covered in our comparison of counting scales vs automated counting.
How an automatic seed counter works
A vision-based seed counter such as the Ball Coleman Seed Counter is a drop-zone counter. Seeds are fed into a drop zone and fall past a high-resolution camera that sees each seed from two views through 90-degree mirrors. Imaging software counts every seed, separates seeds that overlap, and measures each one as it passes.
| Seed type | Example | Counting rate |
|---|---|---|
| Small round seed | 1.25 mm pellets | Up to 2,000 seeds per second |
| Oblong seed | Marigold | 200–300 seeds per second |
| Large seed | Squash and corn | 80–100 seeds per second |
At 2,000 seeds per second, a 50,000-seed sample is counted in about 25 seconds. In repeat testing, two counts of the same 50,000-seed sample varied by only 0.16%. Because each seed is imaged, the counter also records length, width and area, and with an optional top-load scale it reports weight per 1,000 seeds and seeds per gram.
What affects seed counting accuracy
- Overlap and touching seeds. Single-beam counters see two touching seeds as one; two camera views let software separate them.
- Seed shape. Oblong and flat seed tumbles as it falls, which is why it counts more slowly than round seed.
- Size range within a lot. Very small and very large seeds in one sample need settings that suit both.
- Debris, chaff and broken seed. Size measurement helps tell whole seed from fragments and dust.
- Coatings and pellets. Pelleted seed is rounder and easier to count; coatings can change how seed flows.
- Feed rate. Feeding faster than the counter can image raises the chance of overlaps, so rate and accuracy are set together.
How to choose a seed counter
| Question | Why it matters |
|---|---|
| What range of seed sizes and shapes will you count? | Determines optics, drop-zone design and settings per seed type |
| How many seeds per sample or packet, and how many samples per day? | Sets the throughput you need and whether loading time matters more than counting speed |
| How exact must each count be? | Packeting and trials usually need exact counts; rough estimates may tolerate weight counting |
| Do you need size data or seed weight? | Length, width, area and thousand-seed weight add quality data to every count |
| How will lots be tracked? | Stored settings per seed type and lot barcodes speed changeover and support traceability |
| Lab, warehouse or packing line? | Affects footprint, operator interface and data export |
Counting is often only the first step. For germination testing, automated seed and seedling inspection measures seedlings in trays objectively and reproducibly, reducing the labor of reading germination tests by eye.
Need fast, exact seed counts?
The Ball Coleman seed counter counts and measures seed at up to 2,000 per second, separates overlapping seed and stores lot data. It is designed, programmed, built and supported in the US.
Frequently asked questions
An automatic seed counter is the fastest exact method. Vision-based drop-zone counters can count small round seed at up to 2,000 seeds per second, so a 50,000-seed sample takes about 25 seconds.
Only as accurate as the average seed weight used. Seed weight varies between lots and varieties, so a fill weight that is right for one lot can be 10% or more over or under for the next.
Vision-based counters that image each seed are very repeatable. On the Ball Coleman Seed Counter, two counts of the same 50,000-seed sample differed by 0.16%.
Thousand-seed weight is the weight of 1,000 seeds of a lot. It is used to convert a target plant population into a seeding rate by weight and to compare seed size between lots.
Yes. Pelleted seed is round and uniform, which makes it one of the fastest seed types to count; small pellets around 1.25 mm can be counted at up to 2,000 per second.
Related articles
- Automated Counting: The Complete Guide to Industrial Counting Systems
- Counting Scales vs. Automated Counting: Which is Right for You?
- Manual Counting vs. Automated Counting: Cost, Errors and How to Switch
- Counting Machines: How to Count Products Faster with Machine Vision
- AI Counting and Sorting: How It Works and When to Use It
- High-Speed Counter or Counting Machine? How to Choose