Update README.md

dendisuhubdy · Aug 3, 2018 · 2197a6f · 2197a6f
1 parent 1095536
commit 2197a6f
Showing 1 changed file with 10 additions and 11 deletions.
diff --git a/tiny_yolov2/README.md b/tiny_yolov2/README.md
@@ -1,34 +1,33 @@
 # Tiny YOLOv2
 
-Download: 
-- release 1.0: https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_1/tiny_yolov2.tar.gz
-- release 1.2: https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_7/tiny_yolov2.tar.gz
-
-Model size: 61 MB
-
 ## Description
 This model is a real-time neural network for object detection that detects 20 different classes. It is made up of 9 convolutional layers and 6 max-pooling layers and is a smaller version of the more complex full [YOLOv2](https://pjreddie.com/darknet/yolov2/) network. 
 
+## Model
+|Model|Checksum|Download (with sample test data)| ONNX version |Opset version|
+|-----|:-------|:-------------------------------|:-------------|:------------|
+|Tiny YOLOv2|[MD5](https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_1/tiny_yolov2-md5.txt)|[60.8 MB](https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_1/tiny_yolov2.tar.gz) |1.0  |1 |
+|     |[MD5](https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_7/tiny_yolov2-md5.txt)|[60.8 MB](https://www.cntk.ai/OnnxModels/tiny_yolov2/opset_7/tiny_yolov2.tar.gz) |1.2  |7 |
+
 ### Paper
 "YOLO9000: Better, Faster, Stronger" [arXiv:1612.08242](https://arxiv.org/pdf/1612.08242.pdf)
 
 ### Dataset
 The Tiny YOLO model was trained on the [Pascal VOC](http://host.robots.ox.ac.uk/pascal/VOC/) dataset.
 
-## Source
+### Source
 The model was converted from a Core ML version of Tiny YOLO using [ONNXMLTools](https://github.com/onnx/onnxmltools). The source code can be found [here](https://github.com/hollance/YOLO-CoreML-MPSNNGraph). The Core ML model in turn was converted from the [original network](https://pjreddie.com/darknet/yolov2/) implemented in Darknet (via intermediate conversion through Keras).
 
-## Model input and output
+## Inference
 ### Input
 shape `(1x3x416x416)`
+### Preprocessing
 ### Output
 shape `(1x125x13x13)`
-### Pre-processing steps
-### Post-processing steps
+### Postprocessing
 The output is a `(125x13x13)` tensor where 13x13 is the number of grid cells that the image gets divided into. Each grid cell corresponds to 125 channels, made up of the 5 bounding boxes predicted by the grid cell and the 25 data elements that describe each bounding box (`5x25=125`). For more information on how to derive the final bounding boxes and their corresponding confidence scores, refer to this [post](http://machinethink.net/blog/object-detection-with-yolo/).
 ### Sample test data
 Sets of sample input and output files are provided in 
-* the .npz format (`test_data_*.npz`). The input is a `(1x3x416x416)` numpy array of a test image from Pascal VOC, while the output is a numpy array of shape `(1x125x13x13)`.
 * serialized protobuf TensorProtos (`.pb`), which are stored in the folders `test_data_set_*/`.
 
 ## License