QuickSRNet
QuickSRNet upsamples an image by a factor of two.
- Tasks
- restoration
- Sizes
- m2 at 64 px
- Install
pip install libreyolo- Support tier
- Inference only, since v1.6.0. Predict, validate and export only. Training features do not apply.
- Licenses
- Code MIT, weights BSD-3-Clause. Commercial use
Install
pip install "libreyolo"Predict
from libreyolo import LibreYOLO, SAMPLE_IMAGE model = LibreYOLO("LibreQuickSRNetm2-restore.pt", device="cpu")result = model(SAMPLE_IMAGE)result.restored.save("upscaled.png")The Medium 2x model preserves native input resolution and doubles each spatial dimension. Validation uses paired low/high-resolution images and reports PSNR and SSIM. ONNX supports dynamic spatial shapes; TorchScript uses a fixed canvas. Training is not supported.
Validate
from libreyolo import LibreYOLO, SAMPLE_IMAGE model = LibreYOLO("LibreQuickSRNetm2-restore.pt", device="cpu")# Paired super-resolution dataset YAML: inputs/val low-res images, targets/val 2x high-res imagesmetrics = model.val(data="path/to/your/restore.yaml", workers=0)print(metrics)Use the dataset format for this task. Validation explains the dataset requirements and returned metrics.
Export
| Task | ONNX | TorchScript | ExecuTorch | TensorRT | OpenVINO | Paddle | MNN | RKNN | ncnn | TFLite | CoreML | Core AI |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Restoration | Restoration to ONNX: supported | Restoration to TorchScript: supported | Restoration to ExecuTorch: supported | Restoration to TensorRT: supported | Restoration to OpenVINO: supported | Restoration to Paddle: not supported | Restoration to MNN: not supported | Restoration to RKNN: not supported | Restoration to ncnn: supported | Restoration to TFLite: not supported | Restoration to CoreML: not supported | Restoration to Core AI: not supported |
from libreyolo import LibreYOLO, SAMPLE_IMAGE model = LibreYOLO("LibreQuickSRNetm2-restore.pt", device="cpu")model.export(format="onnx")Export setup lists format dependencies and loading exported artifacts.
Checkpoints
| File | Weights license |
|---|---|
| Restoration | |
| LibreQuickSRNetm2-restore.pt | bsd-3-clause |
Every file above exists in the LibreYOLO org today and downloads on first use.
Licensing
Check the license on the Hugging Face repository of the specific weights you download. Every checkpoint in the LibreYOLO org carries one, and they are not always the same across a family. That repository is the authoritative source; the summary below describes what applied when this page was last verified.
This is a description of the licenses involved, not legal advice. If the answer matters commercially, read the licenses yourself and take your own counsel.
- Original work
- QuickSRNet, Qualcomm
- Upstream license
- BSD-3-Clause
- Upstream source
- github.com/quic/aimet-model-zoo
- LibreYOLO code
- MIT
- Weights
- BSD-3-Clause, republished at huggingface.co/LibreYOLO
- Interpretation
- The source and released weights retain the publisher's BSD-3-Clause terms.
