AU-ENG-CVML2018 Identification of Aquatic Macroinvertebrates
- 0 Collaborators
Project of the course "Computer Vision and Machine Learning", in the Department of Engineering at Aarhus University. ...learn more
Project status: Concept
Overview / Usage
This year's project has as theme the identification of Aquatic Macroinvertebrates (or simply bugs in lakes). The types and numbers of macroinvertebrates found in a water body reflect water quality. Thus, this information is used to assess, preserve and assure good water quality.
In this project, a recently collected image dataset, collected within the ongoing project DETECT, is used. A slight modification of the data partitions made publicly available by the project is provided in two forms:
Images (after being cropped to focus on the target object)
Image representations obtained by training a CNN on the training images
Training, validation and test splits are provided, in order to easily compare the performance of the various methods. The training images/representations should be used for training the method, the hyper-parameters of which will be determined by using the validation images/representations. The performance of the trained model, will be subsequently evaluated on the test images/representations using the classification rate metric.
Methodology / Approach
Keras will be used as a front end for Tensorflow to
Enhance data
Transfer learn via bottleneck features
Fine-tune network
Technologies Used
Various machine learning techniques, mostly experiments using CNNs