raab/efishSignalDetector
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

README with ToDos implemented

Browse Source
This commit is contained in:
Till Raab 2023-10-26 11:59:56 +02:00
parent 4bbf3607e9
commit 5c2fc32cea
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
README.md
View File

@ -50,7 +50,6 @@ im2 = ImageOps.grayscale(im1)
* check other pretrained models from torchvision.models.detection, e.g. fasterrcnn_resnet50_fpn_v2 * check other pretrained models from torchvision.models.detection, e.g. fasterrcnn_resnet50_fpn_v2
## config.py ## config.py
Containes Hyperparameters used by the scripts. Containes Hyperparameters used by the scripts.
## ToDos: ## ToDos:
@ -60,6 +59,8 @@ Containes Hyperparameters used by the scripts.
Classes and functions to save models and store loss values for later illustration. Classes and functions to save models and store loss values for later illustration.
Also includes helper functions... Also includes helper functions...
## ToDos:
### train.py ### train.py
Code training the model using the stored images in ./data/dataset and the .csv files Code training the model using the stored images in ./data/dataset and the .csv files
containing the bounding boxes meant for training. For each epoch test-loss (without containing the bounding boxes meant for training. For each epoch test-loss (without
@ -68,9 +69,13 @@ of the previous epochs. If the new model is the best model, the model.state_dict
./model_outputs as best_model.pth. ./model_outputs as best_model.pth.
## ToDos: ## ToDos:
*
### inference.py ### inference.py
Currently, this code performs predictions based in the test dataset (img and corresponding csv file).
However, this code shall be used to infer totally unknown images. Prediction results are ilustrated
and stored in ./inference_output
## ToDo:
* implement path where no csv file is needed...