Marwa H.
Computer Vision Project
Experience
Computer Vision Project
- Developed a Convolutional Neural Network (ConvNet)-based model that achieved 95% accuracy in traffic sign recognition and classification.
- Programming language: Python 3.7
- Libraries: Numpy, matplotlib, scikit-learn, scikit-image
- Deep Learning framework: Tensorflow
NLP Project
- Created and compared the performance of three different models for patent application classification. The first two models use different word representations, while the third is based on a pre-trained language model.
- Accuracy improved significantly—from 60% with the baseline model (using GloVe embeddings) to 88% with the pre-trained model.
- Programming language: Python 3.7
- Word representation: GloVe word embeddings, pre-trained domain-specific embeddings
- Pre-trained language model: Google AI's BERT
- Python’s libraries: Pandas, Numpy
- Deep Learning framework: Tensorflow
Passau, Germany
Master’s thesis at the University of Passau
University of Passau
- Depth information is extracted from RGB images using Generative Adversarial Networks (GANs). The generator follows an encoder-decoder architecture. A pre-trained EfficientNet model is used as the encoder backbone for feature extraction, while the decoder is implemented as a fully convolutional neural network.
- In the first approach, Conditional GANs (CGANs) are trained in a supervised manner using single RGB images paired with corresponding ground truth depth maps for both training and testing.
- The second approach employs a GAN-based architecture trained in an unsupervised fashion using stereo image pairs without ground truth depth maps; at test time only single RGB images with their corresponding depth maps are used.
- Developed using the PyTorch framework.
Languages
Arabic
NativeEnglish
AdvancedFrench
AdvancedGerman
IntermediateEducation
Oct 2018 - Feb 2022
University of Passau
Master's Degree · Computer Science · Passau, Germany
Oct 2014 - Aug 2017
École Pluridisciplinaire Internationale Sousse
Engineering Degree · Electrical Engineering · Sousse, Tunisia
Sep 2011 - Jun 2014
Institut Supérieur d'Informatique et Mathématiques
Bachelor's degree · Electronics and Communication Technology · Monastir, Tunisia
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.