Benchmarking and Validation

[matthew-d-campbell]

Evaluation of model performance against subsetting training data, or subsets of ground-truthed data sets provides information on model performance, however another step beyond that is necessary for implementation of pipelines to derive data sets, estimate abundance, and compare performance. This includes both model precision as well as process efficiency (i.e. does this save us time).

[matthew-d-campbell]

• Create validation test sets by pipeline
• Import bulk video data
• Process time-series
• Estimate indices of abundance
• Estimate processing time
• Create toolset and package