This repository showcases the power of pre-trained TensorFlow models for image classification tasks using popular datasets: Deep Weeds, EuroSAT, and SVHN (Street View House Numbers). Utilizing Streamlit, a Python library for building interactive web applications, this project provides a user-friendly interface for uploading images and obtaining predictions from the models.
- Deep Weeds: Classifies plants into different categories (e.g., Chinee apple, Lantana, etc.).
- EuroSAT: Identifies land use and land cover classes from satellite images (e.g., AnnualCrop, Forest, etc.).
- SVHN Cropped: Recognizes cropped digits from street view images (digits 0-9).
-
Clone the Repository:
git clone https://github.com/saidislombek-abdusamatov/tensorflow_datasets.git
-
Install Dependencies:
pip install tensorflow streamlit opencv-python pandas numpy Pillow
-
Run the Streamlit App:
streamlit run app.py
This command will start the Streamlit web application locally. You can access the app in your browser at
http://localhost:8501.
-
Upload Image:
- Click on the "Upload Image" button to select an image (PNG, JPG, or JPEG format).
- The uploaded image will be displayed below the button.
-
Select Model:
- Choose the model you want to use from the dropdown menu: Deep Weeds, EuroSAT, or SVHN Cropped.
-
Prediction:
- Click the "Predict!" button to see the model's prediction based on the uploaded image.
- The predicted class label will be displayed along with a bar chart showing the model's confidence scores for each class.
- Deep Weeds Classes: Chinee apple, Lantana, Parkinsonia, Parthenium, Prickly acacia, Rubber vine, Siam weed, Snake weed, Negative
- EuroSAT Classes: AnnualCrop, Forest, HerbaceousVegetation, Highway, Industrial, Pasture, PermanentCrop, Residential, River, SeaLake
- SVHN Cropped Classes: Digits 0-9