Deep learning - CentraleSupélec - SDI-M (2024-2025)

Lectures

• Lecture 1 Introduction to deep learning and linear networks
• Lecture 2 Linear networks and RBF networks
• Lecture 3 Convolutional Neural Networks
• Lecture 4 Convolutional Neural Networks : Applications
• Lecture 5 Recurrent Neural Networks and applications

Labs

• Lab 1 Introduction à pytorch
• Lab 2 Segmentation sémantique avec un modèle encodeur/décodeur sur Pascal-VOC
• Lab 3 Automatic Speech Recognition
• Lab 4 Deep Convolutional Generative Adversarial Networks

Challenges

• 2024-2025 Zooplankton semantic segmentation

To have an idea about the performances of the teams, below are renderings of the predictions on the test images where the color code is :

• blue: true negatives
• green : true positives
• red : false positives
• purple: false negatives

🥱 F1=0.926

C'est lui ☝️, F1=0.92364

AST, F1=0.91470

العبرة بالخواتيم, F1=0.91287

Random predictions 2, F1=0.91151

Huguo Bui, F1=0.90933

C'est qui l'patron ? 🙄, F1=0.90897

Eliott, F1=0.90218

., F1=0.90061

100% Homemade UNet, F1=0.88907

TN, F1=0.88580

Still in the bottom, F1=0.86602

Wassim Behlaj, F1=0.85713

Etienne Andrier, F1=0.83777

• 2023-2024 : we participated to the GeoLifeClef 2022 challenge, after the competition was over. The final leaderboard of the teams is shown below.

• 2022-2023 Phytoplankton temporal series prediction
• 2021-2022 Zooplankton classification