| lat_mean=39.951853021309034 | |
| lat_std=0.0005336591548786815 | |
| lon_mean=-75.19122782343982 | |
| lon_std=0.0004335530163099028 | |
| --- | |
| license: mit | |
| --- | |
| #ResNet GPS Prediction Model | |
| This model predicts GPS coordinates (latitude and longitude) from input images using a custom ResNet-based architecture. | |
| ##How to Use | |
| 1. Download `resnet_gps_model.pth` and `config.json` from this repository. | |
| 2. Define the model architecture (as shown in the usage example below). | |
| 3. Load the model weights and configuration. | |
| ## Example Usage | |
| ```python | |
| import torch | |
| import json | |
| # Load config | |
| config = json.load(open("config.json", "r")) | |
| # Define and load model | |
| resnet = CustomResNetModel(model_name="microsoft/resnet-18", num_classes=config["num_classes"]) | |
| state_dict = torch.load("resnet_gps_model.pth") | |
| resnet.load_state_dict(state_dict) | |
| resnet.eval() | |
| This is our customresnetmodel | |
| class CustomResNetModel(nn.Module): | |
| def __init__(self, model_name="microsoft/resnet-18", num_classes=2): | |
| super(CustomResNetModel, self).__init__() | |
| # Load pre-trained ResNet from Hugging Face | |
| self.resnet = AutoModelForImageClassification.from_pretrained(model_name) | |
| # Adjust the classifier layer to output the desired number of classes | |
| in_features = self.resnet.classifier[0].in_features # Assuming the last layer is a Linear layer | |
| self.resnet.classifier = nn.Sequential( | |
| nn.Flatten(), | |
| nn.Linear(in_features, num_classes) | |
| ) | |
| def forward(self, x): | |
| return self.resnet(x) | |
| def save_model(self, save_path): | |
| """Save model locally using the Hugging Face format.""" | |
| self.save_pretrained(save_path) | |
| def push_model(self, repo_name): | |
| """Push the model to the Hugging Face Hub.""" | |
| self.push_to_hub(repo_name) | |