ash001/cats-dogs-transferlearning-cnn

🐱🐶 Cats vs Dogs Classification (Transfer Learning CNN | PyTorch)

A PyTorch image classifier for Cats vs Dogs, trained with transfer learning using a pretrained ResNet-18 backbone.

This Hugging Face repo hosts the trained weights + small config files used by the Streamlit inference app.

Training → Model → Inference

• Training notebook (Colab): https://colab.research.google.com/drive/1FA2K2TimczQBrYaLipwjXpU0BBuw6YDB
• Inference app (Streamlit): https://github.com/sparklerz/Deep-Learning-Fundamentals-Suite
  (page: pages/06_Cats_vs_Dogs_CNN.py)

What’s in this repo

Stored under artifacts/:

• model_state.pt — PyTorch state_dict for ResNet-18 (final FC trained)
• config.json — image size + normalization stats
• class_names.json — class labels (index order)

Model details (high level)

• Backbone: ResNet-18 (ImageNet pretrained)
• Training: backbone frozen, only the final fully-connected layer trained
• Input size: 224×224 RGB
• Normalization: ImageNet mean/std (stored in artifacts/config.json)

Inputs

• One RGB image (any size). At inference time:
  • resize to 256
  • center-crop to 224
  • convert to tensor
  • normalize with mean/std from config.json

Output

• Predicted label: one of ["cats", "dogs"] (from class_names.json)
• You can also compute probabilities via softmax.

Quickstart (load + predict)

import json
import torch
import torch.nn as nn
from PIL import Image
from torchvision import models, transforms
from huggingface_hub import hf_hub_download

REPO_ID = "ash001/cats-dogs-transferlearning-cnn"

# Download artifacts
cfg_path   = hf_hub_download(REPO_ID, "artifacts/config.json")
names_path = hf_hub_download(REPO_ID, "artifacts/class_names.json")
pt_path    = hf_hub_download(REPO_ID, "artifacts/model_state.pt")

config = json.load(open(cfg_path, "r"))
class_names = json.load(open(names_path, "r"))

# Rebuild model architecture + load weights
model = models.resnet18(weights=None)
model.fc = nn.Linear(model.fc.in_features, int(config["num_classes"]))
state = torch.load(pt_path, map_location="cpu")
model.load_state_dict(state)
model.eval()

# Preprocess
tfm = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(int(config["image_size"])),
    transforms.ToTensor(),
    transforms.Normalize(mean=config["mean"], std=config["std"]),
])

img = Image.open("your_image.jpg").convert("RGB")
x = tfm(img).unsqueeze(0)

with torch.no_grad():
    logits = model(x)
    probs = torch.softmax(logits, dim=1)[0].tolist()

pred_idx = int(torch.argmax(logits, dim=1).item())
print("Prediction:", class_names[pred_idx])
print("Probabilities:", dict(zip(class_names, probs)))

---

license: apache-2.0