Trento University Artificial Intelligence Department Deep Learning Course Project
- Prof: Elisa Ricci
- Teaching Assistant: Willi Menapace
In this assignment, we demonstrate the usage of neural networks to solve two common computer vision tasks using the ResNet50 pre-trained model by PyTorch framework. Market-1051 dataset is video-surveillance dataset containing images of multiple persons each of which is captured multiple times by different cameras along with a set of annotations that specify attributes of each person such as age, gender and clothing. The first part of the assignment, we have done multi-class classifier task to predict attributes for each image. In the second part of the assignment, we tried to solve a person re-identification problem using triplet loss.
Students
- Ali Akay
- Mert Akkor
pytorch 1.9.0+cu102
You can find the code also in Google COLAB
We have done hyperparameter Tuning for classification part. You can find the code also in Google COLAB
I used triplet loss for the person re-identification problem. You can find the code also in Google COLAB
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.