Dataset Directory Structure

The directory structure of the dataset should look like,

Root
│   
│      
│
└──Vehicle_Damage_Detection_Dataset
   │   
   │   
   │
   └───annotations
   │   
   │   
   └───images
   

Use of the Repository

1. Install the dependencies after cloning the repo by running the command pip install -r requirements.txt.
2. To perform the training, run the command, python main.py --train.
3. To perform the testing, run the command, python main.py --evaluate.
4. The endpoint for the model inference has been defined inside the inference directory.
5. To run the model endpoint locally run, python main.py --serve.

The outputs are saved in the output directory along with visualizations

To test the API created for the model endpoint

Description

This endpoint accepts an image file and returns predictions for objects detected in the image. The predictions include bounding boxes, labels, and confidence scores for each detected object.

Request

• URL:[(https://damage-detection-447027078795.us-central1.run.app/predict/)]
• Method: POST
• Content-Type: multipart/form-data

file: The image file to be analyzed. This should be provided as a form-data file upload.

Example Request using Windows Powershell

curl.exe -X POST "https://damage-detection-447027078795.us-central1.run.app/predict/" -F "file=@path_to_your_image.jpg"

Response

• Content-Type: application/json
• Body: predictions: A list of prediction objects, each containing: box: A list of four float values representing the coordinates of the bounding box [x_min, y_min, x_max, y_max]. label: The label of the detected object. score: The confidence score of the prediction.

Example Response

{
    "predictions": [
        {
            "box": [
                121.8379898071289,
                146.30760192871094,
                638.38134765625,
                490.4845886230469
            ],
            "label": "severe-broken",
            "score": 0.5068358778953552
        }
    ]
}

Notes

• The endpoint uses a pre-trained model to perform object detection on the provided image.
• Only predictions with a confidence score greater than 0.5 are included in the response.