Ensure you have the following packages installed:
Flask: A lightweight web framework for building web applications in Python.
NumPy: A powerful library for numerical computing with support for multi-dimensional arrays and matrices.
Pandas: A versatile library for data manipulation and analysis.
TensorFlow (tf): An open-source machine learning framework for building and training models.
OpenCV (cv2): A library for computer vision and image processing tasks.
To run the proram locally , just locate in the directory and run the command
python Api.py
MobileNetV2 is a convolutional neural network architecture specifically designed for efficient deep learning inference on mobile and embedded devices. It serves as an ideal backbone network for object detection tasks, offering a balance between accuracy and computational efficiency, making it suitable for deployment in resource-constrained environments.
Key Features: Efficient Architecture: MobileNetV2 introduces inverted residual blocks and linear bottlenecks to optimize computational resources, allowing for efficient feature extraction.
Inverted Residuals: Lightweight depthwise convolutions followed by pointwise convolutions with a bottleneck layer enable more efficient use of computational resources.
Linear Bottlenecks: By employing linear activations in bottleneck layers, MobileNetV2 reduces computational cost and improves gradient flow during training.
Width Multiplier and Resolution Multiplier: These hyperparameters offer flexibility to adjust the model's size and computational complexity according to the requirements of the target device or application.
Used Gunicorn and apache2 on Ubuntu 24.04 to hostthe app on a local server (http://192.168.86.23 )
Please note that if the site is unresponsive, I might have killed the server to save battery power and storage of local laptop. So please just send me a mail(f20211826@goa.bits-pilani.ac.in) if the problem is persisting and I will re initiate the server whenever needed.
A user can expect a huge JSON format array after the process which will contain relevant information about the image used in object detection.