SilageSnap is available on Google Play and iTunes!
It is important to calibrate the app for each camera to ensure accurate results. Arrange at least three pennies or dimes around a quarter on a dark background, ensuring they are not touching. Hold the cameral level with the background and avoid strong lighting that could cause a glare. Capture images starting at a height of approximately 1 ft above the background. Compare results with the know diameter of a penny (19.1 mm) or dime (17.9 mm). Continue to move the camera closer until the results are within 1 mm. Use the height from the background that resulted in a less than a 1 mm difference for all future images.
- Place 1 quarter and at least 3 pennies or dimes on a black, matte surface
- Avoid an area with bright lights that could cause a glare
- Hold the camera level with the background
- Capture images, starting with the camera 1 ft above the background
- Capture images at closer distances until the size predicted by the app matches the known size of a penny (19.1 mm) or dime (17.9) within 1 mm
- Record the final camera height for future image collection
The following are examples of both good and bad calibration images and their results.
Figure 1. Example calibration image using a quarter and pennies.
The results from Figure 1 demonstrate the calibration is complete. The app detected 5 particles, the same number of coins in the image. The mean particle diameter was 19.23 mm which is 0.1 mm different than the expected diameter of a penny (19.1 mm).
Particles smaller than 4.75mm : 0.00%
Number of particles : 5
Average area mm^2: 1226.71
Standard deviation of area mm^2: 11.67
Average diameter mm: 19.23
Standard deviation of diameter mm: 0.10
Figure 2. Example of a bad calibration image. The dust particles affect calibration results. The number of particles increases and the average diameter decreases.
Particles smaller than 4.75mm : 0.79%
Number of particles : 34
Average area : 176.07
Standard deviation of area : 433.59
Average diameter : 3.00
Standard deviation of diameter : 6.74
Collect a representative forage sample (about 1-2 handfuls), at least once per field. It would be best to collect several handfuls, mix them, and then collect 1-2 handfuls from the larger sample.
The kernels must be separated from stover before images can be captured. Water separation is an effective, simple method for separation. See http://fyi.uwex.edu/forage/making-sure-your-kernel-processor-is-doing-its-job/for a more detailed description.
- Fill a dishpan (or similarly sized container) about ¾ full of water
- Place the representative forage sample in the container
- Gently stir the material, for about a minute, to separate the stover from the kernel. The stover will float while the kernel will sink.
- Skim the stover from the surface using your hands or a strainer
- Slowly pour the water from the dishpan to ensure the kernels remain in the dishpan
- Remove any remaining plant material from the sample by hand
- Carefully soak up excess water using a paper towel (the kernels will be less likely to clump if they have some water removed)
- Use a dark, matte background. Black construction paper works well.
- Spread the kernels in a single layer, without touching onto the construction paper. It may require several pieces of paper to fit all the kernels. A paintbrush can aid in spreading out kernels.
- Hold the camera level with the background.
- Hold camera at the height determined during the calibration process
- Avoid cutting off kernels at the edge of the images
- Avoid bright lights that could cause glare.
Figure 3. Example of an acceptable image. None of the kernels are cut off at the edges and are well spaced.
Figure 4. Example of a bad image, multiple kernels are cut off in the frame.
Figure 5. Example of a bad image, kernels are clumped too closely.
Figure 6. Example of bad image due to shiny background and app results. Using black construction paper helps to avoid this issue.
- Particles smaller than 4.75 mm is equivalent to the KPS
- KPS will be lower for wet samples than dry samples (add 6 points to the app results to get the equivalent sieve results)
- Scores higher than 70 indicate optimally processed kernels
- Scores between 50 and 69 indicate adequately processed kernels
- Scores below 50 indicate inadequately processed kernels
- Score will be lower for wet samples than dry samples
- Decreasing the kernel processer roll gap can improve kernel processing