Code and gaze data for CVPR17 paper
We provide the follwoing data for three different datasets (CUB-VW, CUB-SW, and PETS)
- Gaze raw data
- Gaze-based class embeddings of three types GH, GFG, and GFS
- Image features extracted from GoogLeNet using Caffe
We provide two projects:
- Gaze data collection experiment written in C# using Tobii TX300 Eye tracker SDK. The experiment steps are described and explained in the paper
- Structured Joint Embedding for Zero-shot learning implementation using python from Evaluation of Output Embeddings for Fine-Grained Image Classification paper