def detect_selective_search(self, image_fnames):
"""
Do windowed detection over Selective Search proposals by extracting
the crop and warping to the input dimensions of the net.
Take
image_fnames: list
Give
detections: list of {filename: image filename, window: crop coordinates,
predictions: prediction vector} dicts.
"""
# Make sure that 'selective_search_ijcv_with_python' is on the python path:
search_root = '/Users/chickenchen/githubrepo/' # this file is expected to be in {caffe_root}/examples
import sys
sys.path.insert(0, search_root)
import selective_search_ijcv_with_python as selective_search
# Make absolute paths so MATLAB can find the files.
image_fnames = [os.path.abspath(f) for f in image_fnames]
windows_list = selective_search.get_windows(
image_fnames,
cmd='selective_search_rcnn'
)
# Run windowed detection on the selective search list.
return self.detect_windows(zip(image_fnames, windows_list))
def _assemble_images_selective_search(image_fnames):
"""
Run Selective Search window proposals on all images, then for each
image-window pair, extract a square crop.
Input:
image_fnames: list
Output:
images_df: pandas.DataFrame
With 'image', 'window', 'filename' columns.
"""
windows_list = selective_search.get_windows(image_fnames)
data = []
for image_fname, windows in zip(image_fnames, windows_list):
image = load_image(image_fname)
for window in windows:
window_image, _ = format_image(image, window, cropped_size=True)
data.append({
'image': window_image[np.newaxis, :],
'window': window,
'filename': image_fname
})
images_df = pd.DataFrame(data)
return images_df
def _assemble_images_selective_search(image_fnames, n_max=-1):
"""
Run Selective Search window proposals on all images, then for each
image-window pair, extract a square crop.
Input:
image_fnames: list
n_max: max number of boxes to mine, -1 means no restriction
Output:
images_df: pandas.DataFrame
With 'image', 'window', 'filename' columns.
"""
windows_list = selective_search.get_windows(image_fnames, n_max)
data = []
nWindows = 0.
for image_fname, windows in zip(image_fnames, windows_list):
image = load_image(image_fname)
print "{}: {:d} windows found".format(image_fname, windows.shape[0])
nWindows += windows.shape[0]
for window in windows:
window_image, _ = format_image(image, window, cropped_size=True)
data.append({
'image': window_image[np.newaxis, :],
'window': window,
'filename': image_fname
})
nImgs = len(image_fnames)
print "mean number of windows found for {:d} images : {:.3f}".format(
nImgs, nWindows / nImgs)
images_df = pd.DataFrame(data)
return images_df
def _assemble_images_selective_search(image_fnames):
"""
Run Selective Search window proposals on all images, then for each
image-window pair, extract a square crop.
Input:
image_fnames: list
Output:
images_df: pandas.DataFrame
With 'image', 'window', 'filename' columns.
"""
windows_list = selective_search.get_windows(image_fnames)
data = []
for image_fname, windows in zip(image_fnames, windows_list):
image = load_image(image_fname)
for window in windows:
window_image, _ = format_image(image, window, cropped_size=True)
data.append({
'image': window_image[np.newaxis, :],
'window': window,
'filename': image_fname
})
images_df = pd.DataFrame(data)
return images_df
def _assemble_images_selective_search(image_fnames):
"""
Run Selective Search window proposals on all images, then for each
image-window pair, extract a square crop.
Input:
image_fnames: list
Output:
images_df: pandas.DataFrame
With 'image', 'filename' columns.
"""
windows_list = selective_search.get_windows(image_fnames)
data = []
for image_fname, windows in zip(image_fnames, windows_list):
image = load_image(image_fname)
for window in windows:
data.append(
{
"image": format_image(image, window, CROPPED_DIM)[np.newaxis, :],
"window": window,
"filename": image_fname,
}
)
images_df = pd.DataFrame(data)
return images_df
def _assemble_images_selective_search(image_fnames, n_max=-1):
"""
Run Selective Search window proposals on all images, then for each
image-window pair, extract a square crop.
Input:
image_fnames: list
n_max: max number of boxes to mine, -1 means no restriction
Output:
images_df: pandas.DataFrame
With 'image', 'window', 'filename' columns.
"""
windows_list = selective_search.get_windows(image_fnames, n_max)
data = []
nWindows = 0.
for image_fname, windows in zip(image_fnames, windows_list):
image = load_image(image_fname)
print "{}: {:d} windows found".format(image_fname, windows.shape[0])
nWindows += windows.shape[0]
for window in windows:
window_image, _ = format_image(image, window, cropped_size=True)
data.append({
'image': window_image[np.newaxis, :],
'window': window,
'filename': image_fname
})
nImgs = len(image_fnames)
print "mean number of windows found for {:d} images : {:.3f}".format(nImgs, nWindows/nImgs)
images_df = pd.DataFrame(data)
return images_df
def detect_selective_search(self, image_fnames):
"""Do windowed detection over Selective Search proposals by extracting the crop and warping to the input dimensions of the net.
Parameters
image_fnames: list
Returns
detections: list of {filename: image filename, window: crop coordinates, predictions: prediction vector} dicts."""
# Make absolute paths so MATLAB can find the files.
image_fnames = [os.path.abspath(f) for f in image_fnames]
windows_list = selective_search.get_windows(image_fnames, cmd='selective_search_rcnn')
# Run windowed detection on the selective search list.
return self.detect_windows(zip(image_fnames, windows_list))
def detect_selective_search(self, image_fnames):
"""
Do windowed detection over Selective Search proposals by extracting
the crop and warping to the input dimensions of the net.
Take
image_fnames: list
Give
detections: list of {filename: image filename, window: crop coordinates,
predictions: prediction vector} dicts.
"""
import selective_search_ijcv_with_python as selective_search
# Make absolute paths so MATLAB can find the files.
image_fnames = [os.path.abspath(f) for f in image_fnames]
windows_list = selective_search.get_windows(image_fnames)
# Run windowed detection on the selective search list.
return self.detect_windows(zip(image_fnames, windows_list))
def detect_selective_search(self, image_fnames):
"""
Do windowed detection over Selective Search proposals by extracting
the crop and warping to the input dimensions of the net.
Take
image_fnames: list
Give
detections: list of {filename: image filename, window: crop coordinates,
predictions: prediction vector} dicts.
"""
print ">>>>>>>>>>>>>>>> Detector detect_selective_search"
import selective_search_ijcv_with_python as selective_search
# Make absolute paths so MATLAB can find the files.
image_fnames = [os.path.abspath(f) for f in image_fnames]
windows_list = selective_search.get_windows(image_fnames)
# Run windowed detection on the selective search list.
return self.detect_windows(zip(image_fnames, windows_list))
def getProposal(imageName):
obj_proposals = selective_search.get_windows({imageName})
return obj_proposals
