Name		Name	Last commit message	Last commit date
Latest commit History 140 Commits
cnns		cnns
data		data
detection		detection
test_data/instagram		test_data/instagram
utils		utils
.gitignore		.gitignore
README.md		README.md
classlabel_tp.txt		classlabel_tp.txt
compute_test_score_gpu.py		compute_test_score_gpu.py
get_probs.py		get_probs.py
make_plots.py		make_plots.py
set_env.sh		set_env.sh
set_env_sushkov_local.sh		set_env_sushkov_local.sh

Repository files navigation

cs231n-project

To run caffe on a new dataset:
- put the JPEG's into a train/ and val/ directories (use split_train_val.py)
- generate a label file for caffe using generate_caffe_path_label.py (on each line: filepath label)
- run create_leveldb.sh (modified version of caffe's create_imagenet.sh) with the appropriate paths for the train and val directories
- if we wish to use the imagenet image means during training, make sure that file has been downloaded (follow caffe instructions)
- if we wish to use the current dataset's image means during training, run make_mean.sh (modified version of caffe's make_imagenet_mean.sh)
- follow the remaining instructions at BVLC/caffe#550

About

No description, website, or topics provided.

Report repository

Releases

No releases published

Packages

No packages published

Languages

Python 72.3%
Shell 27.7%