ResNet18 Intel Image Classifier

Three fine-tuned ResNet18 models trained on the Intel Image Classification dataset (buildings, forest, glacier, mountain, sea, street), comparing different transfer learning strategies.

Files

resnet18_feature_extract.pth — only the final layer trained (frozen backbone)
resnet18_partial.pth — layer3, layer4, and final layer trained
resnet18_full.pth — entire network fine-tuned

Test accuracy

Strategy	Test Accuracy
feature_extract	87.97%
partial	94.07%
full	93.60%

Usage

import torch
from torchvision import models
import torch.nn as nn

model = models.resnet18(weights=None)
model.fc = nn.Linear(model.fc.in_features, 6)
model.load_state_dict(torch.load("resnet18_partial.pth", map_location="cpu"))
model.eval()

Classes: buildings, forest, glacier, mountain, sea, street

Downloads last month: -; Downloads are not tracked for this model. How to track