Simple but stronge baseline for binarized super-resolution networks (BSRNs).
See the original paper E2FIF: Push the limit of Binarized Deep Imagery Super-resolution using End-to-end Full-precision Information Flow.
This repo is modified from EDSR.
To train and reproduce the results of the paper, just run "train.sh/train_rcan.sh/train_rdn.sh".
Modify the params in "train.sh", like "model", "save", "binary_model" and so on. Then, '''shell sh train.sh '''
By Bolt.
Step 1. Prepare your binarized onnx model. It should be noted that the BN layer will be fused with Conv when pytorch is converted to onnx, which may destroy the binarized conv layers.
Step 2. See the Start pape for Bolt and select your target platform and build platform, like: '''shell ./install.sh --target=android-aarch64 --gpu '''
Step 3. Copy the compiled X2bolt, benchmark, and your onnx models to your phone by adb.
Step 4. Test the real latency of your models.