RNN
01- RNN_Classification
Simple RNN training for classification task of 3 signal: Sine, Square, Triangle.
02- RNN_Regression
Simple RNN training for sine wave estimation.
03- RNN_vs_GRU_Classification
Comparison of RNN model and GRU model for classification task of 3 signal: Sine, Square and Triangle, after 100 epoch training.
Model | Accuracy |
---|---|
RNN Model | 0.9315 |
GRU Model | 0.9383 |
04- RNN_vs_GRU_Regression
Comparison of RNN model and GRU model for regression task of sine wave estimation after 100 epoch training.
Model | loss |
---|---|
RNN Model | 0.0027 |
GRU Model | 0.0026 |
05- Ball_Move_Data_Generation
Generate data for ball move direction
06- GRU_Implementation_from_Scratch
GRU implementation from scratch + inference
07-LSTM_Implementation_from_Scrat
LSTM implementation from scratch + inference
08- Ball_Move_Direction_Classification
-
Generate data for ball move direction
-
Classification of direction using RNN, GRU and LSTM
09- VideoClassificationCRNN
- 09- Video_Classification_CRNN.ipynb(train)
- inference.py
- models.py (gru, lstm, rnn)
- load_video.py
- requirements.txt
Model
Backbone: ResNet50V2
and my vgg base model
for feature extraction
RNN modules: RNN, GRU and LSTM are tested
The performance of GRU module was better than other madules
Dataset
Dataset contains videos from 2 classes
Due to insufficient data, the training was not done well. but this project can be used for other video classification tasks using CRNNs
10- Video_Classification_CRNN
- Video classificatio nusing CRNN on ucf101_top5 dataset
Model
Backbone: my vgg base model
for feature extraction
RNN modules: RNN, GRU are tested
The performance of GRU module was better than RNN madules
Dataset
- ucf101_top5 dataset containing 573 video from 5 classes
Result
Model | Val Accuracy |
---|---|
RNN Model | 0.87 |
GRU Model | 0.94 |