| """Standalone inference for the Tiny Vial CNN Hugging Face release.""" |
|
|
| from __future__ import annotations |
|
|
| import argparse |
| from pathlib import Path |
|
|
| import numpy as np |
| from PIL import Image, ImageDraw |
| import torch |
| from safetensors.torch import load_file |
| from torch import nn |
|
|
|
|
| IMAGE_SIZE = 64 |
|
|
|
|
| class TinyVialDetector(nn.Module): |
| def __init__(self) -> None: |
| super().__init__() |
| self.features = nn.Sequential( |
| nn.Conv2d(3, 12, 3, stride=2, padding=1), nn.ReLU(), |
| nn.Conv2d(12, 24, 3, stride=2, padding=1), nn.ReLU(), |
| nn.Conv2d(24, 32, 3, stride=2, padding=1), nn.ReLU(), |
| ) |
| self.head = nn.Sequential( |
| nn.Flatten(), |
| nn.Linear(32 * 8 * 8, 64), nn.ReLU(), |
| nn.Linear(64, 4), nn.Sigmoid(), |
| ) |
|
|
| def forward(self, image: torch.Tensor) -> torch.Tensor: |
| return self.head(self.features(image)) |
|
|
|
|
| def predict(image_path: Path, weights_path: Path) -> tuple[int, int, int, int]: |
| model = TinyVialDetector() |
| model.load_state_dict(load_file(weights_path, device="cpu")) |
| model.eval() |
| with Image.open(image_path) as source: |
| width, height = source.size |
| resized = source.convert("RGB").resize((IMAGE_SIZE, IMAGE_SIZE), Image.Resampling.BILINEAR) |
| array = np.asarray(resized, dtype=np.float32).copy() |
| tensor = torch.from_numpy(array).permute(2, 0, 1).unsqueeze(0) / 255.0 |
| with torch.inference_mode(): |
| cx, cy, box_width, box_height = model(tensor)[0].tolist() |
| x0 = max(0, min(width - 1, round((cx - box_width / 2) * width))) |
| y0 = max(0, min(height - 1, round((cy - box_height / 2) * height))) |
| x1 = max(x0, min(width - 1, round((cx + box_width / 2) * width) - 1)) |
| y1 = max(y0, min(height - 1, round((cy + box_height / 2) * height) - 1)) |
| return x0, y0, x1, y1 |
|
|
|
|
| def main() -> None: |
| parser = argparse.ArgumentParser() |
| parser.add_argument("image", type=Path) |
| parser.add_argument("--weights", type=Path, default=Path("model.safetensors")) |
| parser.add_argument("--output", type=Path, default=Path("prediction.png")) |
| args = parser.parse_args() |
| bbox = predict(args.image, args.weights) |
| image = Image.open(args.image).convert("RGB") |
| ImageDraw.Draw(image).rectangle(bbox, outline="#76b900", width=2) |
| image.save(args.output) |
| print({"bbox_2d_tight": list(bbox), "output": str(args.output)}) |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
