SAGE Sky Images and model checkpoints for "Let's Unleash the Network Judgement: A Self-supervised Approach for Cloud Image Analysis" paper

This dataset contains:

84912 images collected from our camera (a 6MP Fisheye Sky facing camera with 192 degrees horizontal and vertical angle view). Each image is a JPEG image, JFIF standard 1.01, aspect ratio, density 1x1, segment length 16, baseline, precision 8, 2048x2048, components 3. Images were recorded on October–December 2021. The dataset is clusterized with images belonging to 20 different clusters/classes. We used PCA_MLE plus SOM to produce such clusterization. Files Cluster_*.zip

11 model checkpoints from the training process. Files checkpoint*.pth and log.txt

Output feature vectors and their corresponding file names resulting from inference . File features.zip

The clusterization produced on WSISEG-Database. File README.txt