40 Real World Data Science, Machine Learning Projects 2026

Create a pancard detector with opencv by resizing images to 250 by 160 and converting them to grayscale. Compare originals with tampered images using structural similarity index and thresholds.

Define an index function to handle get and post requests for image upload, resize, grayscale, contours, and bounding rectangles, with a prediction rendered on the index page.

Deploy a Python app to Heroku by configuring the environment and installing dependencies from requirements.txt. Push code to GitHub, connect the repo, then deploy to obtain the app URL.

Test the deployed pan card detector by uploading tampered images to verify a 28.57% correct tamper detection, while deploying a computer vision app with a front end, GitHub, and Heroku.

Build, train, and test a convolutional neural network to identify dog breeds with Keras and TensorFlow in Google Colab, covering Kaggle API setup, data download, preprocessing, and CNN architecture.

Preprocess the dog image dataset by loading labels, selecting breeds (Scottish Deerhound, Maltese dog, Bernese Mountain Dog), one-hot encoding targets, converting images to numpy arrays, and normalizing pixels by 255.

Build a sequential cnn with conv2d and max pooling, flatten, and dense layers for three classes with softmax; train with categorical cross-entropy and adam at 0.0001 learning rate, 100 epochs.

Plot model and validation accuracy by epoch with matplotlib, save the history, and use the predict function to reach 83.33% test accuracy while tuning hyperparameters.

Create a requirements file, install streamlit, navigate directories in the anaconda prompt, run the app with streamlit run main_app.py, and view the dog breed prediction on localhost:8501.

Learn to add image and text watermarks with OpenCV in Google Colab, including loading assets, resizing, region of interest, and merging the watermark for content protection.

Use Google Colab to apply image watermarking with OpenCV, adding logo and text watermarks for copyright protection, and load images and logos via URL and PIL in this project.

Learn to create text and image watermarks using OpenCV and numpy. Convert images to RGB, compute center and ROI, and apply a logo or text watermark for copyright protection.

Deploy a Python app to Heroku by connecting GitHub, creating a new app, enabling automatic deploys from main, installing requirements, and testing a watermarking front end.

Visualize a traffic sign dataset and resize images to 50 by 50; normalize by 255, convert to numpy arrays, and one-hot encode 43 classes with an 80/20 split.

set up an OCR workflow in google colab with tesseract, pytesseract, and opencv. download and install tesseract, import required libraries, fetch an internet image, and extract text.

Learn how to extract text from images using pytesseract with custom configurations, filter out unwanted symbols, and refine results with grayscale conversion, noise reduction, and thresholding in OpenCV.

Create a flask ocr app by configuring project structure, app wiring, and image upload to extract text; set up front end, requirements, tesseract, and deployment with guncon.

Install the project requirements with pip install -r requirements.txt, then run python app.py, open the app in a browser, and test OCR by selecting an image to extract text.

Build a convolutional neural network model by designing conv 2d and max pooling layers, a flatten and dense stack, and softmax for a three-class classifier, then train, evaluate, and save.

Detect and count vehicles in images and video using OpenCV and Haar cascades, loading images via requests, converting to grayscale, and drawing contours with rectangles for cars and buses.

Convert images to grayscale, apply gaussian blur to reduce noise, dilate and perform morphological processing, then detect cars and buses with haar cascades, and output a processed video with counts.

Build a flask app for vehicle detection and counting using OpenCV cascade classifier, image preprocessing, and post-processing to draw rectangles. Deploy via Heroku using GitHub, requirements.txt, and a proc file.

Import and initialize OpenCV, Dlib, NumPy, requests, and PIL in Google Colab, download the 68 face landmarks shape predictor, and load source and destination images from URLs for face extraction.

Convert images to numpy arrays and grayscale, detect faces with dlib, predict landmarks, triangulate, warp triangles, and use seamless clone to finalize the face swap.

Create a Flask-based face swap app by building project structure, configuring settings, and implementing views for image upload and processing. Design the front end with HTML, CSS, and materialize.

Learn to build a convolutional neural network in Keras on Google Colab, loading a bird species dataset from Google Drive, and train a multiclass classifier with 224 by 224 images.

Create and run a Flask app for a six-class bird species classifier, loading a pre-trained model, configuring development settings, handling image uploads, preprocessing, and presenting predictions via a web interface.