Merge pull request #299 from yuyu2172/notice-eval

Notify users about evaluation code
chainer · Jul 3, 2017 · d77a25d · d77a25d
2 parents 1422e24 + d47656e
commit d77a25d
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 0 deletions.
diff --git a/examples/detection/README.md b/examples/detection/README.md
@@ -31,3 +31,12 @@ You can generate these visualization results by the following command.
 ```
 $ python visualuze_models.py
 ```
+
+## Notes on writing your own evaluation code
+
+Here is a list of important configurations to reproduce results.
+
++ `model.use_preset('evaluate')` configures postprocessing parameters for evaluation such as threshold for confidence score.
++ `DetectionVOCEvaluator` should be instantiated with `use_07_metric=True` (default is False), if evaluation is conducted on VOC 2007 test dataset.
++ When evaluating on VOC dataset, `VOCDetectionDataset` should return information about difficulties of bounding boxes, as the evaluation metric expects that to be included.
+The dataset returns it by setting `use_difficult=True` and `return_difficult=True`.
diff --git a/examples/faster_rcnn/README.md b/examples/faster_rcnn/README.md
@@ -52,6 +52,12 @@ $ python train.py [--gpu <gpu>]
 ```
 
 
+### Evaluation
+
+The evaluation score is reported by `DetectionVOCEvaluator` during training.
+Also, the evaluation can be conducted outside of training loop by using [`chainercv/examples/detection/eval_voc07.py`](https://github.com/chainer/chainercv/blob/master/examples/detection).
+
+
 ### References
 This code is based on Caffe implementation by the original authors https://github.com/rbgirshick/py-faster-rcnn and Chainer a re-implementation https://github.com/mitmul/chainer-faster-rcnn .
 

diff --git a/examples/ssd/README.md b/examples/ssd/README.md
@@ -22,6 +22,11 @@ Convert `*.caffemodel` to `*.npz`. Some layers are renamed to fit ChainerCV. SSD
 $ python caffe2npz <source>.caffemodel <target>.npz
 ```
 
+## Evaluation
+
+The evaluation can be conducted using [`chainercv/examples/detection/eval_voc07.py`](https://github.com/chainer/chainercv/blob/master/examples/detection).
+
+
 ## References
 1. Wei Liu, et al. "SSD: Single shot multibox detector" ECCV 2016.
 2. Cheng-Yang Fu, et al. "[DSSD : Deconvolutional Single Shot Detector](https://arxiv.org/abs/1701.06659)" arXiv 2017.