Towards Scalable and Accurate Fine-Grained Categorization /

In this work I contribute two solutions towards increasing the utility and performance of fine-grained categorization. First, I present a software infrastructure designed to ease the burden of collecting and managing a fine-grained datasets. Second, I present a technique that significantly advances state-of-the-art performance on bird species categorization. These contributions provide the groundwork for expanding to other fine-grained domain