First commit

.dockerignore

@@ -0,0 +1,9 @@
+mlops
+model_funcs
+__pycache__
+*.pyc
+*.log
+.git
+.venv
+*.swp
+*.swo

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+app/cd.pt filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,21 @@
+FROM python:3.12.3
+
+# Create non-root user with UID 1000
+RUN useradd -m -u 1000 user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+WORKDIR /app
+
+# Copy requirements and install
+COPY --chown=user requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy app code
+COPY --chown=user app/ app/
+
+# Expose port 7860 (used by HF Spaces)
+EXPOSE 7860
+
+# Run FastAPI with uvicorn on the required port
+CMD ["python3", "-m", "uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -1,10 +1,32 @@
 ---
-title: Testing Spaces
-emoji: 👀
+title: Cat vs Dog Classifier 🐶🐱
+emoji: 🐾
 colorFrom: gray
 colorTo: red
 sdk: docker
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🐾 Cat vs Dog Classifier - FastAPI + PyTorch
+
+This Space hosts a FastAPI-based image classification API that distinguishes between **cats** and **dogs** using a fine-tuned **ResNet50** model.
+
+## 🚀 How It Works
+
+- Accepts an uploaded image via a POST request.
+- Processes the image using `albumentations` and a pre-trained PyTorch model.
+- Returns `"cat"` or `"dog"` based on prediction confidence.
+
+## 🧪 Try It Out
+
+You can interact with the `/predict-image` endpoint using tools like:
+- `curl`
+- Postman
+- Any frontend that supports `multipart/form-data` file uploads.
+
+### Example:
+
+```bash
+curl -X POST \
+  -F "file=@your_image.jpg" \
+  https://Klasta-testing_spaces.hf.space/predict-image

app/cd.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba6880fed2ae00203357b8de47b92f31d1d7ad4fb6b5c45b79968e47328aa583
+size 94343841

app/main.py

@@ -0,0 +1,26 @@
+import fastapi
+import io
+from fastapi import UploadFile, File
+from PIL import Image
+from .predict import predict_image
+
+app = fastapi.FastAPI()
+
+@app.get("/")
+def root():
+    return {"message": "API is working!"}
+
+@app.post("/predict-image")
+async def upload_image(file: UploadFile = File(...)):
+    if not file.content_type.startswith("image/"):
+        return {"error": "Please upload an image to proceed"}
+
+    contents = await file.read()
+
+    try:
+        image = Image.open(io.BytesIO(contents))
+    except Exception as e:
+        return {"error": f"Failed to open image: {str(e)}"}
+
+    pred = predict_image(image)
+    return {"prediction": pred}

app/model.py

@@ -0,0 +1,22 @@
+import torch
+import torch.nn as nn
+import torchvision.models as models
+
+
+class CatvsDogResNet50(nn.Module):
+    def __init__(self, freeze_backbone: bool = True):
+        super().__init__()
+        self.backbone = models.resnet50(pretrained=True)
+
+        if freeze_backbone:
+            for param in self.backbone.parameters():
+                param.requires_grad = False
+
+        num_ftrs = self.backbone.fc.in_features
+        self.backbone.fc = nn.Sequential(
+            nn.Dropout(0.5),
+            nn.Linear(num_ftrs, 1),
+        )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.backbone(x)

app/predict.py

@@ -0,0 +1,34 @@
+# app/predict.py
+
+import torch
+import numpy as np
+from PIL import Image
+import albumentations as A
+from albumentations.pytorch import ToTensorV2
+from .model import CatvsDogResNet50
+
+transform = A.Compose(
+    [
+        A.Resize(224, 224),
+        A.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)),
+        ToTensorV2(),
+    ]
+)
+
+device = "cuda" if torch.cuda.is_available() else "cpu"  # Fallback to CPU if no GPU
+model = CatvsDogResNet50(freeze_backbone=True)
+model.load_state_dict(torch.load(r"app/cd.pt", map_location=device)) 
+model = model.to(device).eval()
+
+
+def predict_image(image: Image.Image) -> str:
+    img = image.convert("RGB")
+    img = np.array(img)
+    img = transform(image=img)["image"]
+    img = img.unsqueeze(0).to(device)
+
+    with torch.inference_mode():
+        out = model(img)
+        prob = torch.sigmoid(out).item()
+        pred = "dog" if prob > 0.5 else "cat"
+    return pred

requirements.txt

@@ -0,0 +1,9 @@
+fastapi==0.115.12
+uvicorn==0.34.3
+torch==2.7.1
+torchvision==0.22.1
+albumentations==2.0.8
+pillow==11.2.1
+numpy==2.1.3
+python-multipart==0.0.20
+mlflow==3.1.0