The flourishing application of drones within marine science provides more opportunity to conduct photogrammetric studies on large and varied populations of many different species. While these new platforms are increasing the size and availability of imagery datasets, established photogrammetry methods require considerable manual input, allowing individual bias in techniques to influence measurements, increasing error and magnifying the time required to apply these techniques. Here, we introduce the next generation of photogrammetry methods utilizing a convolutional neural network to demonstrate the potential of a deep learning‐based photogrammetry system for automatic species identification and measurement. We then present the same data analysed using conventional techniques to validate our automatic methods. Our results compare favorably across both techniques, correctly predicting whale species with 98% accuracy (57/58) for humpback whales, minke whales, and blue whales. Ninety percent of automated length measurements were within 5% of manual measurements, providing sufficient resolution to inform morphometric studies and establish size classes of whales automatically. The results of this study indicate that deep learning techniques applied to survey programs that collect large archives of imagery may help researchers and managers move quickly past analytical bottlenecks and provide more time for abundance estimation, distributional research, and ecological assessments.

CNN Test Imagery and LabelsThis zip file includes the test images used for reported metrics along with the via_region_data.json label file that has the mask shapes in JSON format as needed for the neural network. Species specific .json files are also included.test.zipCNN Training Imagery and Labels Subset 1This zip file includes the first subset of training images used for training this CNN. This data is only subset for ease of upload and download but should be combined by the user on their local machine. The full label file for all training data is included with both subsets and is identical. That file "via_region_data.json" has the mask shapes in and species IDs in a JSON format as needed for the neural network.train_subset_1.zipCNN Training Imagery and Labels Subset 2This zip file includes the second subset of training images used for training this CNN. This data is only subset for ease of upload and download but should be combined by the user on their local machine. The full label file for all training data is included with both subsets and is identical. That file "via_region_data.json" has the mask shapes in and species IDs in a JSON format as needed for the neural network.train_subset_2.zipCNN Validation Imagery and LabelsThis zip file includes the validation images used for training this CNN. The label file for all validation data is included. That file "via_region_data.json" has the mask shapes in and species IDs in a JSON format as needed for the neural network.val.zip