README.md

@@ -8,7 +8,7 @@
 
 AIR is a deep learning based object detection solution to automate the aerial drone footage inspection task frequently carried out during search and rescue (SAR) operations with drone units. It provides a fast, convenient and reliable way to augment aerial, high-resolution image inspection for clues about human presence by highlighting relevant image regions with bounding boxes, as done in the image below. With the assistance of AIR, SAR missions with aerial drone searches can likely be carried out much faster before, and with considerably higher success rate.
 
-This code repository is based on the master's thesis work by **Pasi Pyrrö** <span itemscope itemtype="https://schema.org/Person"><a itemprop="sameAs" content="https://orcid.org/0000-0002-0277-4996" href="https://orcid.org/0000-0002-0277-4996" target="orcid.widget" rel="me noopener noreferrer" style="vertical-align:top;"><img src="https://orcid.org/sites/default/files/images/orcid_16x16.png" style="width:1em;margin-right:.5em;" alt="ORCID iD icon"></a></span> from [Aalto University, School of Science](https://www.aalto.fi/en/school-of-science) with the title "AIR: Aerial Inspection RetinaNet for Land Search and Rescue Missions". This thesis work was funded by Accenture.
+This code repository is based on the [master's thesis work](http://urn.fi/URN:NBN:fi:aalto-202202061749) by **Pasi Pyrrö** <span itemscope itemtype="https://schema.org/Person"><a itemprop="sameAs" content="https://orcid.org/0000-0002-0277-4996" href="https://orcid.org/0000-0002-0277-4996" target="orcid.widget" rel="me noopener noreferrer" style="vertical-align:top;"><img src="https://orcid.org/sites/default/files/images/orcid_16x16.png" style="width:1em;margin-right:.5em;" alt="ORCID iD icon"></a></span> from [Aalto University, School of Science](https://www.aalto.fi/en/school-of-science) which was funded by [Accenture](https://www.accenture.com/fi-en/about/company/finland).
 
 <img src="keras_retinanet/images/air-example.png" alt="AIR-example"/><br><br>
 

evaluate.sh

@@ -19,6 +19,7 @@ MODEL="${3:-$MODEL}"
 DATASET="${4:-$DATASET}"
 
 python3 keras_retinanet/keras_retinanet/bin/evaluate.py \
+    --gpu 0 \
     --backbone resnet152 \
     --image_min_side 1525 \
     --image_max_side 2025 \

infer.sh

@@ -16,6 +16,7 @@ MODEL="${2:-$MODEL}"
 DATASET="${3:-$DATASET}"
 
 python3 keras_retinanet/keras_retinanet/bin/infer.py \
+    --gpu 0 \
     --backbone resnet152 \
     --image_min_side 1525 \
     --image_max_side 2025 \

requirements.txt

@@ -5,7 +5,7 @@ opencv-python==4.5.3.56
 scikit-learn
 filterpy==1.4.5
 Pillow
-tensorflow==1.15.0
+tensorflow==1.15.5
 keras
 keras-resnet==0.1.0
 wandb==0.9.5

setup.py

@@ -54,7 +54,7 @@ def run(self):
 
 setuptools.setup(
     name             = 'air-detector',
-    version          = '0.0.1',
+    version          = '0.0.2',
     description      = 'Keras implementation of Aerial Inspection RetinaNet object detector.',
     url              = 'https://github.com/Accenture/AIR',
     author           = 'Pasi Pyrrö',
@@ -71,8 +71,8 @@ def run(self):
                         'cython', 'Pillow', 'opencv-python', 'progressbar2', 
                         'wandb==0.9.5', 'scikit-learn', 'matplotlib', 'filterpy'],
     extras_require   = {
-        "cpu":        ["tensorflow==1.15.0"],
-        "gpu":        ["tensorflow-gpu==1.15.0"],
+        "cpu":        ["tensorflow==1.15.5"],
+        "gpu":        ["tensorflow-gpu==1.15.5"],
     },
     entry_points     = {
         'console_scripts': [

train.sh

@@ -19,6 +19,7 @@ BACKBONE="${3:-$BACKBONE}"
 STEPS=3
 
 python3 keras_retinanet/keras_retinanet/bin/train.py \
+    --gpu 0 \
     --random_transform=true \
     --anchor_scale=0.965 \
     --config=$PWD/keras_retinanet/config.ini \