Fish Metadata
A note on the metadata CSV file
This model integrates descriptive metadata for each fish with scale images to predict age. The current version of the model uses fish length, weight, and the month the fish was caught. These variables were found, through trial and error, to be the most helpful attributes for improving age predictions. The model requires this information be stored in a single CSV file with one entry per sample to be processed. This CSV file can be called anything, but for this discussion, we will call it metadata.csv for convenience.
Consider three images of scales named 25123.jpg, 25234.jpg, and 25345.jpg, stored in a single directory, for which ages are desired. The accompanying metadata.csv file would look like this:
| image_name | length | weight | month |
|---|---|---|---|
| 25123.jpg | 208 | 147 | 6 |
| 25234.jpg | 219 | 176 | 5 |
| 25345.jpg | 222 | 183 | 6 |
| … | … | … | … |
| 26315.jpg | 211 | 168 | 5 |
Note that the column headers can be anything, but the model expects these variables in this order. Also notice that each image file has an entry and each entry is on its own line. If an image lacks a metadata entry, or if a metadata entry exists but an image does not, the model will fail to run. Finally, the image file names in the first column do not need to include the file extension. If the extensions are missing, they will automatically be added using the output_type extension provided in configurations.yml.
The column headers in metadata.csv can be called anything, but the variables must be provided in the order shown: image name, length, weight, month of catch.