This dataset is used to compressive sensing scenes and assist users in training the optimal sparsity threshold. The input features are image compression ratio, image size, and sparsity, while the label field is the optimal sparsity threshold. The test images are from the USC-SIPI dataset.