Skip to content

SaifurRR/DeepLearning-ComputerVision-ImageProcessing

Repository files navigation

DeepLearning-NeuralNet-ComputerVision-ImageProcessing

Neural Network:

Digit Recognizer (Kaggle MNIST dataset)

Dataset: https://www.kaggle.com/c/digit-recognizer

1. Cross Entropy Loss using Logistic Regression for Single-Layer NN as a Benchmark

Final error rate: $0.08$

ALT TEXT

Visual Perception Autonomous Driving:

1. Stereo Depth to Driving Scenario

ALT TEXT
Closest point depth (meters): $28.92655372619629$

2. Visual Odometry for Localization in Autonomous Driving

ALT TEXT

Figure: Camera Trajectory(3D Visualization)

3. Environment Perception For Self Driving Cars

ALT TEXT

Figure: Estimate distance to impact

4. Mission Planning Dijkstras & A* search Algorithm

ALT TEXT

Computer Vision (Kaggle):

1. Convolutional Classifier

ALT TEXT

after 30 epochs: val_loss: 0.3322, val_binary_accuracy: 0.8620