CNN Playground

Interactive Deep Learning Exploration

Explore convolutional neural networks through hands-on experimentation. Upload images, draw digits, detect objects, and understand how deep learning works in real-time.

IMG

ImageNet Classification

Upload any image and see how ResNet50 classifies it among 1000 categories. Explore Grad-CAM attention maps to understand where the network focuses, and visualize feature activations across different layers.

ResNet50 Classification Grad-CAM Visualization Feature Map Explorer Top-K Predictions
Upload Image →
DRAW

MNIST Digit Recognition

Draw digits (0-9) on an interactive canvas and watch a trained CNN predict in real-time. See confidence scores for each digit and understand how the network processes your handwriting with live preprocessing visualization.

Interactive Drawing Canvas Real-time Prediction Mnist_Net 99.66% Accuracy
Start Drawing →
VISION

Computer Vision

Advanced object detection and segmentation using state-of-the-art models. Choose between image analysis, video processing, or live camera detection with both server-side PyTorch and client-side TensorFlow.js models.

YOLO Object Detection Instance Segmentation Video Processing Live Camera Detection
Detect Objects →

Technology Stack

Backend & API

Python 3.8+ FastAPI PyTorch OpenCV

Frontend & UI

HTML5 & CSS3 JavaScript Canvas API WebRTC

Neural Networks

YOLO11 ResNet50 Custom CNN TensorFlow.js

Computer Vision

Object Detection Image Segmentation Real-time Analysis Video Processing