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

README update

Browse Source
This commit is contained in:
Till Raab 2023-10-26 12:09:16 +02:00
parent 214a4bd10c
commit 16bca48cbe
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
README.md
View File

@ -3,8 +3,8 @@
Welcome to the efishSignalDetector, a **neural network** framework adapted
for detecting electrocommunication signals in wavetype electric fish based on
spectrogram images. The model itself is a pretrained **FasterRCNN** Model with a
**ResNet50** Backbone. Only the final predictor is replaced to not predict 91 classes
the coco-dataset it is trained to but the (currently) 1 category it should detect.
**ResNet50** Backbone. Only the final predictor is replaced to not predict the 91 classes
present in the coco-dataset the model is trained to but the (currently) 1 category it should detect.
## Data preparation
### Data structure
@ -29,7 +29,9 @@ training and one for testing (both also stored in ./data/dataset).
## model.py
Contains the model and adjustements.
Contains the detection neural network model with its adjustments. The model itself is a pretrained **FasterRCNN** Model with a
**ResNet50** Backbone. Only the final predictor is replaced to not predict the 91 classes
present in the coco-dataset the model is trained to but the (currently) 1 category it should detect.
### ToDos:
* replace backbone entry to not take RGB input, but grayscale images or even spectrograms.
@ -54,6 +56,7 @@ Containes Hyperparameters used by the scripts.
### ToDos:
* replace TRAIN_DIR with DATA_DIR everywhere !!!
* Do we need the RESIZE_TO parameter ?
## custom_utils.py
Classes and functions to save models and store loss values for later illustration.