Data Collection

Sampling approaches, camera settings and tips for in-the-field collection of photographic data to be used in ReefCloud

How many photos do I need?

Point count sampling approaches can provide relatively precise benthic cover estimates for groups that are not overly rare. The amount of sampling required, in terms of both the number of images and number of points per image, will vary with the abundance, size and distribution of target biota. Generally for broad-scale monitoring work increasing the number of photos (rather than the number of points per photo) improves precision (see Perkins et al 2016).

A good general rule is to aim for at least 20-30 photos per transect. If you are establishing a new monitoring program, it will be important to consider the effects of size, abundance and patterns of distribution of your target biota on the precision of estimates, and adjust your monitoring design accordingly.

Do I need to use a special camera?

The better quality the photos, the better data you will have available to extract, but any underwater camera will do! It is more important to maintain consistency in the camera and settings used throughout your project.

What camera settings should I use?

Most introductory camera users will benefit from accepting the "Auto" or "Underwater" mode settings, if available. These settings will adjust for differences in lighting at depth and balance the colours captured in each photo.

More experienced camera users which have the use of a camera with full manual mode may collect higher quality images by adjusting the aperture, shutter speed and ISO settings given the light availability and conditions. You may want to consider adjusting some of the settings prior to being underwater and a number of online guides are available to assist.

The most important consideration is to remain consistent in your use of settings as much as possible throughout your project.

Top tips for taking underwater photos for benthic monitoring:

1. Keep the camera as still as possible when taking your image to improve resolution and reduce blurriness.

2. Face the camera directly downwards to the benthos, rather than along the reef or at an angle. Filling the frame with will allow you to capture a good representation of the benthic cover.

3. Try to maintain a consistent distance from the bottom, capturing an area of reef between 30 - 150 cm. The closer you are, the better your colour, contrast and sharpness will be; however remember consistency is important, so make sure the camera is the same distance above the reef for each photo in your Project. A great tip for maintaining the same distance from the bottom to the camera is to use a plastic quadrat, a fishing weight on a line of known length, or your extended arm.

4. When creating a new Project in ReefCloud, you will specify the 'field of view' for the model to consider. The 'field of view' provides an estimate of how much reef area you think your photos consistently capture.

Some examples of images that will work well in ReefCloud, and some that are problematic due to colour imbalance, inconsistent field of view, angle taken, poor photo quality and a few photo bombers.

Is it a problem if my tape or quadrat are in the frame of an image?

Not a problem! Just remember to add an image label that is appropriate, such as “tape” or “quadrat” to your labelset so if an annotation point falls on that object the AI model can be taught to recognise it. You can map these labels to the Benthic Category 'Other' when carrying out your Label Mapping. Alternatively, images can be disabled from surveys so that fins, faces and other poor quality images do not get included in in the analysis of benthic cover.

What file size is best for my images?

It is best to try and keep image sizes below 10MB. There was previously a restriction of 6MB per image, but based on feedback we now compress images further to display in ReefCloud.