def predict_tile(self,
raster_path=None,
image=None,
patch_size=400,
patch_overlap=0.05,
iou_threshold=0.15,
return_plot=False,
use_soft_nms=False,
sigma=0.5,
thresh=0.001):
"""For images too large to input into the model, predict_tile cuts the
image into overlapping windows, predicts trees on each window and
reassambles into a single array.
Args:
raster_path: Path to image on disk
image (array): Numpy image array in BGR channel order
following openCV convention
patch_size: patch size default400,
patch_overlap: patch overlap default 0.15,
iou_threshold: Minimum iou overlap among predictions between
windows to be suppressed. Defaults to 0.5.
Lower values suppress more boxes at edges.
return_plot: Should the image be returned with the predictions drawn?
use_soft_nms: whether to perform Gaussian Soft NMS or not, if false, default perform NMS.
sigma: variance of Gaussian function used in Gaussian Soft NMS
thresh: the score thresh used to filter bboxes after soft-nms performed
Returns:
boxes (array): if return_plot, an image.
Otherwise a numpy array of predicted bounding boxes, scores and labels
"""
self.model.eval()
result = predict.predict_tile(model=self.model,
raster_path=raster_path,
image=image,
patch_size=patch_size,
patch_overlap=patch_overlap,
iou_threshold=iou_threshold,
return_plot=return_plot,
use_soft_nms=use_soft_nms,
sigma=sigma,
thresh=thresh,
device=self.device)
#edge case, if no boxes predictioned return None
if result is None:
print("No predictions made, returning None")
return None
#Set labels to character from numeric if returning boxes df
if not return_plot:
result["label"] = result.label.apply(
lambda x: self.numeric_to_label_dict[x])
return result
def predict_tile(self,
raster_path=None,
image=None,
patch_size=400,
patch_overlap=0.05,
iou_threshold=0.15,
return_plot=False,
mosaic=True,
use_soft_nms=False,
sigma=0.5,
thresh=0.001,
color=None,
thickness=1):
"""For images too large to input into the model, predict_tile cuts the
image into overlapping windows, predicts trees on each window and
reassambles into a single array.
Args:
raster_path: Path to image on disk
image (array): Numpy image array in BGR channel order
following openCV convention
patch_size: patch size default400,
patch_overlap: patch overlap default 0.15,
iou_threshold: Minimum iou overlap among predictions between
windows to be suppressed. Defaults to 0.5.
Lower values suppress more boxes at edges.
return_plot: Should the image be returned with the predictions drawn?
mosaic: Return a single prediction dataframe (True) or a tuple of image crops and predictions (False)
use_soft_nms: whether to perform Gaussian Soft NMS or not, if false, default perform NMS.
sigma: variance of Gaussian function used in Gaussian Soft NMS
thresh: the score thresh used to filter bboxes after soft-nms performed
color: color of the bounding box as a tuple of BGR color, e.g. orange annotations is (0, 165, 255)
thickness: thickness of the rectangle border line in px
Returns:
boxes (array): if return_plot, an image.
Otherwise a numpy array of predicted bounding boxes, scores and labels
"""
# Load on GPU is available
if torch.cuda.is_available():
self.model = self.model.to("cuda")
self.model.eval()
self.model.score_thresh = self.config["score_thresh"]
self.model.nms_thresh = self.config["nms_thresh"]
result = predict.predict_tile(model=self.model,
raster_path=raster_path,
image=image,
patch_size=patch_size,
patch_overlap=patch_overlap,
iou_threshold=iou_threshold,
return_plot=return_plot,
mosaic=mosaic,
use_soft_nms=use_soft_nms,
sigma=sigma,
thresh=thresh,
device=self.current_device,
color=color,
thickness=thickness)
#edge case, if no boxes predictioned return None
if result is None:
print("No predictions made, returning None")
return None
#Set labels to character from numeric if returning boxes df
if not return_plot:
if mosaic:
result["label"] = result.label.apply(lambda x: self.numeric_to_label_dict[x])
else:
for df,image in result:
df["label"] = df.label.apply(lambda x: self.numeric_to_label_dict[x])
return result