Keras implementation of Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. originally from https://github.com/yhenon/keras-frcnn/
-
Annotations and Image data can be found in the data directory.
-
Code file : Faster_RCNN.ipynb
-
Install pre-requisites
$ pip install -r requirements.txt
-
Scripts for creating annotations are included in the code.
-
Results get stored in the results_imgs folder created in the main directory itself. Results of the implemented models can be found here
-
mAP implementation is present in main.py file, originally by Cartucho
-
mAP calculation makes use of ground-truth and detection-results from the input directory. Detection Results are automatically added by test_frcnn.py.