Use softmax scores by default

experiments/test_net.py

@@ -4,9 +4,9 @@ import fast_rcnn_test
 import caffe
 
 if __name__ == '__main__':
-    prototxt = 'model-defs/vgg16_pyramid_forward_only_bbox_reg.prototxt'
-    caffemodel = '../fast-rcnn-bbox04/snapshots/vgg16_finetune_bbox04_bbox06_iter_40000.caffemodel'
-    GPU_ID = 3
+    prototxt = 'model-defs/vgg16_bbox_reg_deploy.prototxt'
+    caffemodel = 'snapshots/vgg16_finetune_joint_flipped_iter_40000.caffemodel'
+    GPU_ID = 2
 
     caffe.set_phase_test()
     caffe.set_mode_gpu()

fast_rcnn_config.py

@@ -45,5 +45,6 @@ SNAPSHOT_ITERS  = 10000
 TEST_SCALES     = (600,)
 TEST_MAX_SIZE   = 1000
 TEST_NMS        = 0.3
+TEST_BINARY     = False
 
 RNG_SEED        = 3

fast_rcnn_test.py

@@ -120,11 +120,12 @@ def _clip_boxes(boxes, im_shape):
 def im_detect(net, im, boxes):
     blobs, im_scale_factors = _get_blobs(im, boxes)
 
-    # TODO: remove duplicates
-    # v = np.array([1, 1e3, 1e6, 1e9, 1e12])
-    # hashes = blobs['rois'][:, :, 0, 0].dot(v.T)
-    hashes = (blobs['rois'][:, :, 0, 0] *
-              np.array([[1, 1e3, 1e6, 1e9, 1e12]])).sum(axis=1)
+    # When mapping from image ROIs to feature map ROIs, there's some aliasing
+    # (some distinct image ROIs get mapped to the same feature ROI).
+    # Here, we identify duplicate feature ROIs, so we only compute features
+    # on the unique subset.
+    v = np.array([1, 1e3, 1e6, 1e9, 1e12])
+    hashes = blobs['rois'][:, :, 0, 0].dot(v.T)
     _, index, inv_index = np.unique(hashes, return_index=True,
                                     return_inverse=True)
     blobs['rois'] = blobs['rois'][index, :, :, :]
@@ -138,9 +139,14 @@ def im_detect(net, im, boxes):
     net.blobs['rois'].reshape(num_rois, 5, 1, 1)
     blobs_out = net.forward(data=blobs['data'].astype(np.float32, copy=False),
                             rois=blobs['rois'].astype(np.float32, copy=False))
-    scores = blobs_out['fc8_pascal'][:, :, 0, 0]
-    # Return scores as fg - bg
-    scores = scores - scores[:, 0][:, np.newaxis]
+    if conf.TEST_BINARY:
+        # simulate binary logistic regression
+        scores = blobs_out['fc8_pascal'][:, :, 0, 0]
+        # Return scores as fg - bg
+        scores = scores - scores[:, 0][:, np.newaxis]
+    else:
+        # use softmax estimated probabilities
+        scores = blobs_out['prob'][:, :, 0, 0]
     box_deltas = blobs_out['fc8_pascal_bbox'][:, :, 0, 0]
     pred_boxes = _bbox_pred(boxes, box_deltas)
     pred_boxes = _clip_boxes(pred_boxes, im.shape)

model-defs/vgg16_pyramid_forward_only_bbox_reg.prototxt → model-defs/vgg16_bbox_reg_deploy.prototxt

@@ -430,6 +430,12 @@ layers {
   weight_decay: 1
   weight_decay: 0
 }
+layers {
+  name: "prob"
+  type: SOFTMAX
+  bottom: "fc8_pascal"
+  top: "prob"
+}
 layers {
   name: "fc8_pascal_bbox"
   type: INNER_PRODUCT