Steps to recreate report results:
- Generate Spectrum images for Histogram Matching:
$ python create_spectrums.py 320 240 4 # Figure 3
- Create simulated copies of training, validation, and test data:
$ mkdir figs $ python match_video.py $ python powerlaw_video.py $ python gen_histograms.py # Figure 1
- Train the CNN models:
$ mkdir models $ python train.py $ python train_gray.py $ python train_spec.py
- Evaluate the CNN models:
$ python eval_models.py # Figure 2 $ python eval_noge_models.py # Figure 4 $ python eval_spec_models.py # Figure 5 $ python eval_ge_models.py # Figure 6
- Evaluate execution timings for each approach
$ python eval_prep_times.py # Table 1