Week 24 - Final results
Summary of the results obtained for modeled and raw images.
Logbook
2020
Week 23 - Modifying training samples number
Analyze the impact of modidy the number of samples in training.
Week 20, 21 & 22 - Training ConvLSTM-4 for raw samples
Train a new LSTM net with different motion types and DOF to improve results.
Week 18 & 19 - Training ConvLSTM for raw samples
Train a ConvLSTM with all motion types and DOF on raw samples.
Week 17 - Pixel extension
Extend the pixel area to try to improve the CNN+LSTM results.
Week 15 & 16 - Training CNN+LSTM for raw samples
Train a combination of CNN with LSTM net with different motion types and DOF.
Week 13 & 14 - Training CNN for raw samples
Train a CNN with all motion types and DOF on raw samples.
Week 12 - Long term prediction
Analyze the impact of increasing the time gap.
Week 11 - Training LSTM-4 for modeled samples
Train a new LSTM net with all motion types and DOF to improve results.
Week 10 - Training LSTM for modeled samples
Train a LSTM with all motion types and DOF on modeled images.
Week 9 - Training MLP for modeled samples
Train a MLP with all motion types and DOF on modeled images.
Week 8 - Generating new data
Generate new frame data using parabolic and sinusoidal dynamics .
Week 7 - Training with modeled frames
Train different types of networks with modeled images.
Week 6 - Refactoring code
Adapt the code to train and evaluate networks with modeled frames.
Week 5 - Training nets with a big dataset
It is evaluated how a large increase in the number of samples affects the efficiency of the networks.
Weeks 3 & 4 - Read the dataset by batches
We focus on adapting the code to be able to read the dataset by batches and solve the memory error obtained with the large dataset.
Week 2 - Training with more raw images
We focus on training using raw images and linear movement with different types of networks and more samples.
Week 1 - Two frame types
New learning is tackled by valuing modeled frames and raw frames.
Week 0 - Checkpoint
Compiling the work done so far and the results obtained to resume the project and establish a new starting point.
Previous weeks - Change in initial height
Linear motion is further complicated by letting the point start at a random height.
Previous weeks - New linear motion
A new degree of freedom is added to the linear motion.
Previous weeks - First trainings
First contact with the created data.
Datasets - Data types
The different types of data used for research in this thesis.