updated submission

Author: werywjw

README.md

@@ -11,6 +11,7 @@
 - [x] requirements.txt ([requirements of packages for GiMeFive](https://github.com/werywjw/SEP-CVDL/blob/main/requirements.txt))
 - [x] README.md ([README for GiMeFive](https://github.com/werywjw/SEP-CVDL/blob/main/README.md))
 - [x] best_GiMeFive.pth ([saved GiMeFive model](https://github.com/werywjw/SEP-CVDL/blob/main/best_GiMeFive.pth))
+- [x] classification_scores_valid.csv ([result from script of classification scores](https://github.com/werywjw/SEP-CVDL/blob/main/classification_scores_valid.csv))
 - [x] script_csv.py ([script that iterates through folders and creates csv file](https://github.com/werywjw/SEP-CVDL/blob/main/script_csv.py))
 - [x] script_csv.sh ([alternative script](https://github.com/werywjw/SEP-CVDL/blob/main/script_csv.sh))
 - [x] script_video.py ([Script that classifies and saves videos and live camera streams](https://github.com/werywjw/SEP-CVDL/blob/main/script_video.py))
@@ -22,13 +23,14 @@
 - [x] plot_gcam.ipynb ([Jupiter Notebook for GradCAM visulization](https://github.com/werywjw/SEP-CVDL/blob/main/plot_gcam.ipynb))
 - [x] plot_heatmap.ipynb ([Jupiter Notebook for Heatmap visulization](https://github.com/werywjw/SEP-CVDL/blob/main/plot_gcam.ipynb))
 - [x] eval_video.ipynb ([Jupiter Notebook that classifies and saves videos and live camera](https://github.com/werywjw/SEP-CVDL/blob/main/eval_video.ipynb))
-- [x] video.mp4 ([test video with no evaluation](https://github.com/werywjw/SEP-CVDL/blob/main/video/video.mp4))
-- [x] eval_video.mp4 ([evaluated video](https://github.com/werywjw/SEP-CVDL/blob/main/eval_video.mp4))
-- [x] haarcascade_frontalface_default.xml ([pretrained model of haar cascade](https://github.com/werywjw/SEP-CVDL/blob/main/haarcascade_frontalface_default.xml))
-- [x] shape_predictor_68_face_landmarks.dat ([pretrained model of Landmarks](https://github.com/werywjw/SEP-CVDL/blob/main/shape_predictor_68_face_landmarks.dat))
 - [x] slurm_jobs.sh ([script for slurm](https://github.com/werywjw/SEP-CVDL/blob/main/slurm/slurm_job.sh))
 - [x] installed_packages.txt ([requirements of packages for slurm](https://github.com/werywjw/SEP-CVDL/blob/main/slurm/installed_packages.txt))
 
+- [ ] video.mp4 ([test video with no evaluation](https://github.com/werywjw/SEP-CVDL/blob/main/video/video.mp4))
+- [ ] eval_video.mp4 ([evaluated video](https://github.com/werywjw/SEP-CVDL/blob/main/eval_video.mp4))
+- [ ] shape_predictor_68_face_landmarks.dat ([pretrained model of Landmarks](https://github.com/werywjw/SEP-CVDL/blob/main/shape_predictor_68_face_landmarks.dat))
+- [ ] haarcascade_frontalface_default.xml ([pretrained model of haar cascade](https://github.com/werywjw/SEP-CVDL/blob/main/haarcascade_frontalface_default.xml))
+
 ## Abstract
 
 Deep convolutional neural networks have been shown to successfully recognize facial emotions for the past years in the realm of computer vision. However, the existing detection approaches are not always reliable or explainable, we here propose our model GiMeFive with interpretations, i.e., via layer activations and gradient-weighted class activation mapping. We compare against the state-of-the-art methods to classify the six facial emotions. Empirical results show that our model outperforms the previous meth- ods in terms of accuracy on two *Facial Emotion Recognition* (FER) benchmarks and our aggregated FER GiMeFive. Furthermore, we explain our work in real-world image and video examples, as well as real-time live streams.