CrabDetect

Installation

git clone https://github.com/CameronBodine/CrabDetect

cd CrabDetect

git clone --depth 1 https://github.com/CameronBodine/PINGMapper

conda env create --file conda/CrabDetect.yml

Minor Tweak to Inference Package

inference uses an API to download models you have trained or uploaded with/to Roboflow. Models are saved in root directory that gets deleted after a restart (/tmp/cache or C:/tmp/cache). This path is stored in a variable MODEL_CACHE_DIR which is set in the miniconda/envs/crabpot/lib/python2.11/site-packages/inference/core/env.py file at line 218:

# Model cache directory, default is "/tmp/cache"
MODEL_CACHE_DIR = os.getenv("MODEL_CACHE_DIR", "/tmp/cache")

Changing that path to somewhere that isn't automatically deleted allows the models to be saved for offline use, such as:

# Model cache directory, default is "/tmp/cache"
# MODEL_CACHE_DIR = os.getenv("MODEL_CACHE_DIR", "/tmp/cache")
MODEL_CACHE_DIR = os.getenv("MODEL_CACHE_DIR", "/home/cbodine/PythonRepos/CrabDetect/Models")

The model needs to be run the first time online, then can be used anytime thereafter without internet connection.