Robot-Visionary-System

This repository contains a script for performing inference using a pre-trained ResNet-50 model. The model is used to classify objects in images based on the ImageNet dataset.

Requirements

Before running the inference script, make sure you have the following dependencies installed:

pip install torch torchvision

Files in this Repository

• objdetection_inference.py: The main inference script.
• resnet50.pth: Pre-trained model weights (ensure this file is present in the directory).
• demo.jpg: Sample image for testing the inference script.

How to Run

1. Prepare the Model and Image

   • Ensure resnet50.pth is in the same directory as the script.
   • Place the test image (e.g., demo.jpg) in the directory.

2. Run the Inference Script

   python objdetection_inference.py

3. Expected Output The script will output the top 5 predicted classes along with their confidence scores. Example output:

   Top Predictions:
   Labrador retriever: 85.2%
   Golden retriever: 7.4%
   Flat-coated retriever: 3.2%
   

Code Explanation

1. Load the Pre-trained Model

   • The script loads the ResNet-50 architecture.
   • The model weights are loaded from resnet50.pth.
   • The model is set to evaluation mode to disable gradient updates.

2. Preprocess the Input Image

   • The image is read using torchvision.io.read_image().
   • Image resizing and normalization are applied using torchvision.transforms.
   • The image is converted into a tensor suitable for ResNet.

3. Perform Inference

   • The processed image is passed through the model.
   • Softmax is applied to get class probabilities.
   • The top 5 predictions are displayed with confidence scores.

Notes

• If you encounter errors with missing model weights, ensure resnet50.pth is downloaded correctly.
• Modify image_path in objdetection_inference.py to test with different images.
• The model is based on ImageNet categories.

License

The ResNet-50 model is provided by PyTorch under its respective license.