VisionLab Image Classification Server

VisionLab is a FastAPI web application for image classification experiments. It serves a small browser interface where users can classify ImageNet sample images, upload their own images, apply basic image transformations, compute grayscale histograms, and download classification results as JSON or PNG plots.

Features

• Classify images with Torchvision models: resnet18, alexnet, vgg16, and inception_v3.
• Download and prepare 1,000 ImageNet sample images plus ImageNet labels.
• Show top-5 classification scores in the browser with Chart.js.
• Upload a local image and classify it without saving the uploaded file on disk.
• Adjust image color, brightness, contrast, and sharpness before classification.
• Compute grayscale image histograms with OpenCV and NumPy.
• Download classification output as JSON or as a generated PNG plot.

Project Structure

.
|-- app/
|   |-- config.py                 # Model list and image resource path
|   |-- prepare_images.py         # Downloads sample images and labels
|   |-- prepare_models.py         # Pre-downloads Torchvision model weights
|   |-- ml/                       # Classification helpers
|   |-- histogram/                # Histogram helpers
|   |-- transformations/          # PIL image transformation helpers
|   |-- forms/                    # Form parsing classes
|   |-- static/                   # Frontend JavaScript and favicon
|   `-- templates/                # Jinja templates
|-- showcase/                     # Static public-facing project overview
|-- main.py                       # FastAPI routes
|-- requirements.txt
`-- Documentation.md              # Original issue implementation notes

Setup

Create and activate a Python environment:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements.txt

Prepare the image resources:

python app/prepare_images.py

Optional: pre-download the model weights before starting the server:

python app/prepare_models.py

Run

Start the FastAPI server:

uvicorn main:app --reload

Open http://127.0.0.1:8000.

Main Routes

Route	Purpose
/	Home page
/info	JSON list of configured models and prepared sample images
/classifications	Select a prepared image and classify it
/classify_transform	Transform a prepared image before classification
/histograms	Compute a grayscale histogram for a prepared image
/uploadImage	Upload and classify a local image
/outputJSON	Download classification scores as JSON
/outputPNG	Download classification scores as a PNG plot

Verification

The app was verified locally with:

python app/prepare_images.py
python -c "import main; print(len(main.app.routes))"
uvicorn main:app --host 127.0.0.1 --port 8001

The following checks returned 200 OK:

GET  /
GET  /info
GET  /classifications
GET  /classify_transform
GET  /histograms
GET  /uploadImage
GET  /docs
POST /histograms
POST /classifications
POST /classify_transform
POST /classifyUpload
GET  /outputJSON
GET  /outputPNG

Showcase Page

The showcase/ folder contains a static project overview page for portfolio or Vercel deployment. If deploying only the showcase on Vercel, set the root directory to showcase/.