Fish Body Condition Tracking

ISEF Category: Animal Sciences

Subcategory: Nutrition and Growth  ·  Difficulty: Intermediate  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

A fish can look healthy in a tank and still lose body condition before you notice by eye. Smartphone photos can turn that hidden change into a number called K, the body condition factor. That gives you a simple way to test whether feeding schedule changes growth signals and energy stores. You are not just taking pictures, you are turning them into data.

What Is It?

Body condition factor, often written as K, is a quick way to describe how heavy a fish is for its length. Think of it like a fit check for fish. A fish that is long but thin will have a lower K than one with the same length and a fuller body. Researchers use that number to get a rough sense of growth, nutrition, and overall health.

In this project, you use smartphone photos to predict K from visible features such as length, body depth, or body area. Regression means building a math model that connects those image features to the real K value. It works a bit like learning how to guess a backpack's weight from how full it looks. The more clean, consistent photos you collect, the better the model can learn the pattern.

Why This Is a Good Topic

This is a strong science fair topic because you can change one clear factor, feeding schedule, and measure a clear response, body condition. The project connects to fish care, aquaculture, and animal nutrition, so the results have real-world value. You can also learn photo standardization, image measurement, and regression analysis without needing a professional lab. That makes the project realistic, testable, and rich enough for original research.

Research Questions

• How does feeding schedule change the predicted body condition factor K of aquarium fish over time?
• What is the effect of using body length alone versus length plus body depth on K prediction accuracy?
• Does a standardized photo setup improve regression accuracy compared with handheld aquarium photos?
• To what extent do tank reflections and background color change image-based K estimates?
• Which image features, such as length, depth, or area, best predict measured K?
• Does a model trained on one feeding schedule still predict K well for fish on a different feeding schedule?

Basic Materials

• Smartphone with a camera and manual exposure controls.
• Clear aquarium or photo tank with a fixed viewing side.
• Calibration ruler or scale marker placed in the photo plane.
• Neutral background sheet or mat for the tank.
• Tripod or phone stand to keep camera angle fixed.
• Fish-safe handling tools approved for your setup.
• Digital scale with gram accuracy for approved weight measurements.
• Measuring board or length board for fish length checks.
• Spreadsheet software on a laptop or tablet for data entry.

Advanced Materials

• Overhead camera rig with fixed distance and lighting.
• Recirculating aquarium system with monitored water quality.
• Precision digital balance for fish mass measurements.
• Measuring board, calipers, and calibration target for image scaling.
• Color calibration card for photo standardization.
• Computer with a Python environment for model building.
• Image segmentation workstation for extracting body shape features.
• Approved fish handling and anesthesia supplies, only if your supervising lab requires them.

Software & Tools

• ImageJ: Measures fish length, area, and body shape features from standardized photos.
• Python: Fits regression models and compares prediction error across feeding groups.
• Google Sheets: Organizes measurements, labels photos, and plots body condition trends.
• RStudio: Runs statistical tests and cross-validation for model checking.
• OpenCV: Helps with image calibration, cropping, and background cleanup in photos.

Experiment Steps

1. Define the fish species, feeding schedules, and exact body condition metric you will predict.
2. Build a photo station with fixed lighting, distance, and a size reference so every image stays comparable.
3. Choose which image features you will measure, such as length, area, or body depth, and decide how you will label each sample.
4. Split your data into training and test sets so you can check whether the model predicts new photos, not just familiar ones.
5. Plan controls for tank background, camera angle, and fish posture, because each one can distort the image signal.
6. Select the error metric you will report, such as mean absolute error or R-squared, and compare it across feeding schedules.

Common Pitfalls

• Photographing fish at different angles, which changes apparent body depth and breaks the regression.
• Letting reflections, bubbles, or glare stay in the frame, which confuses segmentation and length estimates.
• Mixing fish from different species or size classes in one model, which hides body-shape differences.
• Changing water quality at the same time as feeding schedule, which makes the nutrition result hard to interpret.
• Training and testing on nearly identical photos, which makes the model look better than it really is.

What Makes This Competitive

A strong project does more than predict a number from a photo. It checks whether the model still works across tanks, camera angles, and feeding schedules, then reports the error clearly. You can compare simple length-based regression with models that add body depth or area, and use cross-validation to show how well the method generalizes. That kind of design turns a photo project into a real measurement study.

Project Variations

• Test whether the same photo-based model works for goldfish, guppies, or zebrafish instead of one species.
• Compare length-only regression with body-area regression to see which predicts K more accurately.
• Measure whether daylight, LED light, or flash changes the quality of image-based K estimates.

Learn More

• NOAA Fisheries educational pages: Search NOAA for fish growth, nutrition, and body condition basics.
• PubMed: Search review articles on fish body condition factor, growth metrics, and aquaculture nutrition.
• USDA National Agricultural Library: Search for aquaculture nutrition and fish growth resources.
• ImageJ Documentation: Find measurement guides for analyzing length, area, and shape in photos.
• OpenCV Documentation: Find tutorials for image calibration, segmentation, and object detection.
• MIT OpenCourseWare: Search for introductory statistics or data analysis material for building and evaluating regression models.