Merge pull request #9 from sjain-stanford/master

SAME padding for maxpool to match with Darknet (tested on yolov3-tiny)

README.md

@@ -15,7 +15,7 @@ Provide optional argument `--training` to generate training graph (uses batch no
 
 ### Object Detection Networks
 
-#### yolo-v2
+#### yolov2
 ```
 python3 main.py \
     --cfg 'data/yolov2.cfg' \
@@ -25,7 +25,17 @@ python3 main.py \
     --gpu 0
 ```
 
-#### yolo-v3
+#### yolov2-tiny
+```
+python3 main.py \
+    --cfg 'data/yolov2-tiny.cfg' \
+    --weights 'data/yolov2-tiny.weights' \
+    --output 'data/' \
+    --prefix 'yolov2-tiny/' \
+    --gpu 0
+```
+
+#### yolov3
 ```
 python3 main.py \
     --cfg 'data/yolov3.cfg' \
@@ -35,6 +45,16 @@ python3 main.py \
     --gpu 0
 ```
 
+#### yolov3-tiny
+```
+python3 main.py \
+    --cfg 'data/yolov3-tiny.cfg' \
+    --weights 'data/yolov3-tiny.weights' \
+    --output 'data/' \
+    --prefix 'yolov3-tiny/' \
+    --gpu 0
+```
+
 ### Image Classification Networks
 
 #### darknet19

main.py

@@ -20,6 +20,7 @@ def parse_net(num_layers, cfg, weights, training=False, const_inits=True, verbos
     cfg_walker = CFGReader(cfg)
     weights_walker = WeightsReader(weights)
     output_index = []
+    num_layers = int(num_layers)
 
     for ith, layer in enumerate(cfg_walker):
         if ith > num_layers and num_layers > 0:

models/yolov2-tiny.cfg

@@ -0,0 +1,139 @@
+[net]
+# Testing
+batch=1
+subdivisions=1
+# Training
+# batch=64
+# subdivisions=2
+width=416
+height=416
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+learning_rate=0.001
+burn_in=1000
+max_batches = 500200
+policy=steps
+steps=400000,450000
+scales=.1,.1
+
+[convolutional]
+batch_normalize=1
+filters=16
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=1
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+###########
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=425
+activation=linear
+
+[region]
+anchors =  0.57273, 0.677385, 1.87446, 2.06253, 3.33843, 5.47434, 7.88282, 3.52778, 9.77052, 9.16828
+bias_match=1
+classes=80
+coords=4
+num=5
+softmax=1
+jitter=.2
+rescore=0
+
+object_scale=5
+noobject_scale=1
+class_scale=1
+coord_scale=1
+
+absolute=1
+thresh = .6
+random=1

models/yolov2-tiny.weights.txt

@@ -0,0 +1,3 @@
+https://pjreddie.com/media/files/yolov2-tiny.weights
+(or)
+https://drive.google.com/uc?export=view&id=1qzPVXk_ZXcqdllbaLMPeCEB0scEFX2cN

models/yolov3-tiny.cfg

@@ -0,0 +1,182 @@
+[net]
+# Testing
+batch=1
+subdivisions=1
+# Training
+# batch=64
+# subdivisions=2
+width=416
+height=416
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+learning_rate=0.001
+burn_in=1000
+max_batches = 500200
+policy=steps
+steps=400000,450000
+scales=.1,.1
+
+[convolutional]
+batch_normalize=1
+filters=16
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=2
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[maxpool]
+size=2
+stride=1
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+###########
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+
+[yolo]
+mask = 3,4,5
+anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
+classes=80
+num=6
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 8
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+[yolo]
+mask = 0,1,2
+anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
+classes=80
+num=6
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1

models/yolov3-tiny.weights.txt

@@ -0,0 +1,3 @@
+https://pjreddie.com/media/files/yolov3-tiny.weights
+(or)
+https://drive.google.com/uc?export=view&id=1V4gI_vA2kvWrwELRA8RbuXJGMDjrWz1N

util/cfg_layer.py

@@ -101,7 +101,8 @@ def cfg_convolutional(B, H, W, C, net, param, weights_walker, stack, output_inde
 def cfg_maxpool(B, H, W, C, net, param, weights_walker, stack, output_index, scope, training, const_inits, verbose):
     pool_args = {
         "pool_size": int(param['size']),
-        "strides": int(param['stride'])
+        "strides": int(param['stride']),
+        "padding": 'same'
     }
 
     net = tf.layers.max_pooling2d(net, name=scope, **pool_args)