This project is based on an article titled "Building a Bird Recognition App and Large Scale Dataset With Citizen Scientists: The Fine Print in Fine-Grained Dataset Collection", which was authored by the SE(3) Computer Vision Group at Cornell University and presented at the Computer Vision and Pattern Recognition (CVPR) Conference in Boston in 2015. They worked with "citizen scientists and domain experts" to develop a "high quality dataset containing 48,562 images of North American birds with 555 categories, part annotations and bounding boxes" (see https://vision.cornell.edu/se3/building-a-bird-recognition-app-and-large-scale-dataset-with-citizen-scientists-the-fine-print-in-fine-grained-dataset-collection/).
Goals include:
- Using Tensorflow to classify images in Python
- Use a Convolutional Neural Network (CNN) to classify images
- Interpret the results of your CNN using Tensorflow
- Use multiple pre-trained binary classifiers to make predictions on a set of images
Data source: CUB 200 (CUB_200_2011.tgz)