Machine Learning for Human Pose Estimate

Essential python model

Numpy for a highly optimized library for numerical operations with a MATLAB-style syntax.
Chainer neural network (deep learning) framework
opencv3.0 OpenCV-Python for solve computer vision problem

Datasets: subset of full FLIC
There are five different approach:
1. Use all the pixels of transformed pictures as inputs and SGD -> python my_scripts/train.py
2. Use all the pixels of transformed pictures as inputs and BGD -> python my_scripts/train.py --mode bgd
3. Use SIFT based on Bag-of-Word algorithm and SGD -> python my_scripts/train_sift.py
4. Use pixels of patch images around the joints and SGD -> python my_scripts/train_bgd.py
Training results are in Notebook

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
models		models
my_scripts		my_scripts
tests		tests
Least_square.ipynb		Least_square.ipynb
Linear_BGD_lr_0.0001.log		Linear_BGD_lr_0.0001.log
README.md		README.md
linear_regression_training_flic.html		linear_regression_training_flic.html
linear_regression_training_flic.ipynb		linear_regression_training_flic.ipynb
linear_regression_training_flic_bgd.ipynb		linear_regression_training_flic_bgd.ipynb
test_draw_loss.ipynb		test_draw_loss.ipynb
test_sift.ipynb		test_sift.ipynb
test_transform.ipynb		test_transform.ipynb
training_results.ipynb		training_results.ipynb