connecting seasoned-lynx

.gitignore

@@ -23,3 +23,8 @@ venv
 
 # HPC output/error jobs
 *.[oe][0-9]*
+
+# seasoned_lynx model
+/feedback/
+/logs/
+tmp_*.png

.gitmodules

@@ -19,3 +19,6 @@
 [submodule "submodules/models/yolo_legend"]
 	path = submodules/models/yolo_legend
 	url = https://github.com/abodeuis/yolo_legend_model
+[submodule "submodules/models/seasoned_lynx"]
+	path = submodules/models/seasoned_lynx
+	url = https://github.com/dou0000/DARPA_Seasoned_Lynx.git

=0.1.9

@@ -0,0 +1,50 @@
+Looking in indexes: https://pypi.org/simple, https://test.pypi.org/simple/
+Collecting git+https://github.com/DARPA-CRITICALMAAS/[email protected]+draft
+  Cloning https://github.com/DARPA-CRITICALMAAS/cdr_schemas.git (to revision v0.2.6+draft) to /tmp/pip-req-build-nu0p80e8
+  Resolved https://github.com/DARPA-CRITICALMAAS/cdr_schemas.git to commit f49d594157b39e1a7ff5d1c1d308a2c07adbb566
+  Installing build dependencies: started
+  Installing build dependencies: finished with status 'done'
+  Getting requirements to build wheel: started
+  Getting requirements to build wheel: finished with status 'done'
+  Preparing metadata (pyproject.toml): started
+  Preparing metadata (pyproject.toml): finished with status 'done'
+Collecting cmaas_utils
+  Downloading https://test-files.pythonhosted.org/packages/84/f8/ddc84026b19a4787fca31b21b3e4abd69fc798788d9ca96454c942c69d58/cmaas_utils-0.1.14-py3-none-any.whl.metadata (7.8 kB)
+Requirement already satisfied: pydantic in ./venv/lib/python3.10/site-packages (2.8.2)
+Requirement already satisfied: geopandas in ./venv/lib/python3.10/site-packages (from cmaas_utils) (1.0.1)
+Requirement already satisfied: numpy in ./venv/lib/python3.10/site-packages (from cmaas_utils) (1.26.4)
+Requirement already satisfied: rasterio in ./venv/lib/python3.10/site-packages (from cmaas_utils) (1.3.10)
+Requirement already satisfied: annotated-types>=0.4.0 in ./venv/lib/python3.10/site-packages (from pydantic) (0.7.0)
+Requirement already satisfied: pydantic-core==2.20.1 in ./venv/lib/python3.10/site-packages (from pydantic) (2.20.1)
+Requirement already satisfied: typing-extensions>=4.6.1 in ./venv/lib/python3.10/site-packages (from pydantic) (4.12.2)
+Requirement already satisfied: pyogrio>=0.7.2 in ./venv/lib/python3.10/site-packages (from geopandas->cmaas_utils) (0.9.0)
+Requirement already satisfied: packaging in ./venv/lib/python3.10/site-packages (from geopandas->cmaas_utils) (24.1)
+Requirement already satisfied: pandas>=1.4.0 in ./venv/lib/python3.10/site-packages (from geopandas->cmaas_utils) (2.2.2)
+Requirement already satisfied: pyproj>=3.3.0 in ./venv/lib/python3.10/site-packages (from geopandas->cmaas_utils) (3.6.1)
+Requirement already satisfied: shapely>=2.0.0 in ./venv/lib/python3.10/site-packages (from geopandas->cmaas_utils) (2.0.6)
+Requirement already satisfied: affine in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (2.4.0)
+Requirement already satisfied: attrs in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (24.2.0)
+Requirement already satisfied: certifi in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (2024.7.4)
+Requirement already satisfied: click>=4.0 in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (8.1.7)
+Requirement already satisfied: cligj>=0.5 in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (0.7.2)
+Requirement already satisfied: snuggs>=1.4.1 in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (1.4.7)
+Requirement already satisfied: click-plugins in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (1.1.1)
+Requirement already satisfied: setuptools in ./venv/lib/python3.10/site-packages (from rasterio->cmaas_utils) (65.5.0)
+Requirement already satisfied: python-dateutil>=2.8.2 in ./venv/lib/python3.10/site-packages (from pandas>=1.4.0->geopandas->cmaas_utils) (2.9.0.post0)
+Requirement already satisfied: pytz>=2020.1 in ./venv/lib/python3.10/site-packages (from pandas>=1.4.0->geopandas->cmaas_utils) (2024.1)
+Requirement already satisfied: tzdata>=2022.7 in ./venv/lib/python3.10/site-packages (from pandas>=1.4.0->geopandas->cmaas_utils) (2024.1)
+Requirement already satisfied: pyparsing>=2.1.6 in ./venv/lib/python3.10/site-packages (from snuggs>=1.4.1->rasterio->cmaas_utils) (3.1.4)
+Requirement already satisfied: six>=1.5 in ./venv/lib/python3.10/site-packages (from python-dateutil>=2.8.2->pandas>=1.4.0->geopandas->cmaas_utils) (1.16.0)
+Downloading https://test-files.pythonhosted.org/packages/84/f8/ddc84026b19a4787fca31b21b3e4abd69fc798788d9ca96454c942c69d58/cmaas_utils-0.1.14-py3-none-any.whl (17 kB)
+Building wheels for collected packages: cdr_schemas
+  Building wheel for cdr_schemas (pyproject.toml): started
+  Building wheel for cdr_schemas (pyproject.toml): finished with status 'done'
+  Created wheel for cdr_schemas: filename=cdr_schemas-0.2.6+draft-py3-none-any.whl size=17254 sha256=112a1c8bee97fd2125aff01dba3992b4c0f093072bac4f95eef46f49de7df9cb
+  Stored in directory: /tmp/pip-ephem-wheel-cache-sgp3fcx0/wheels/72/58/31/75d1ae21f832e4bca4b96ed0f27bc2c9320e1fe0b5045698e4
+Successfully built cdr_schemas
+Installing collected packages: cdr_schemas, cmaas_utils
+  Attempting uninstall: cdr_schemas
+    Found existing installation: cdr_schemas 0.4.2
+    Uninstalling cdr_schemas-0.4.2:
+      Successfully uninstalled cdr_schemas-0.4.2
+Successfully installed cdr_schemas-0.2.6+draft cmaas_utils-0.1.14

CHANGELOG.md

@@ -6,6 +6,9 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 
 ## [Unreleased]
 
+## [0.4.6] - 2024-08-28
+- Adding seasoned-lynx model.
+
 ## [0.4.5] - 2024-08-20
 
 ### Changed

pipeline.py

@@ -22,12 +22,13 @@
     'blaring_foundry',
     'flat_iceberg',
     'drab_volcano',
-    'icy_resin'
+    'icy_resin',
+    'seasoned_lynx',
 ]
 
 # Lazy load only the model we are going to use
 from src.models.pipeline_model import pipeline_model
-def load_pipeline_model(model_dir : str, model_name : str, override_batch_size=None) -> pipeline_model :
+def load_pipeline_model(model_dir : str, model_name : str, override_batch_size=None, override_patch_size=None) -> pipeline_model :
     """Utility function to only import and load the model we are going to use. Returns loaded model"""
     log.info(f'Loading model {model_name}')
     model_stime = time()
@@ -52,6 +53,13 @@ def load_pipeline_model(model_dir : str, model_name : str, override_batch_size=N
     if model_name == 'icy_resin':
         from src.models.icy_resin_model import icy_resin_model
         model = icy_resin_model()
+    if model_name == 'seasoned_lynx':
+        # from src.models.seasoned_lynx_model import seasoned_lynx_model
+        from src.models.seasoned_lynx_model_recolor import seasoned_lynx_model
+        patch_size = 1024 if override_patch_size is None else override_patch_size
+        model = seasoned_lynx_model(patch_size)
+
+
     if not os.path.exists(os.path.join(model_dir, model._checkpoint)):
         msg = f'Could not load {model_name}. Model checkpoint file "{model._checkpoint}" not found in model directory "{model_dir}"'
         log.error(msg)
@@ -249,6 +257,11 @@ def post_parse_output_types(data : List[str]) -> List[str]:
                         type=str,
                         default=os.getenv("CDR_SYSTEM_VERSION", "0.3.0"),
                         help='The version of the cdr system that will be used to generate cdrs. Default is 1.0')
+    # for seasoned_lynx
+    optional_args.add_argument('--patch_size',
+                        type=int,
+                        default=None,
+                        help='Option to override the default patch size for the seasoned lynx model. Default is 1024')
 
     # Flags
     flag_group = parser.add_argument_group('Flags', '')
@@ -409,7 +422,7 @@ def construct_pipeline(args):
     p = pipeline_manager()
     legend_model = yolo_legend_model()
     legend_model.load_model(args.checkpoint_dir)
-    segmentation_model = load_pipeline_model(args.checkpoint_dir, args.model, override_batch_size=args.batch_size)
+    segmentation_model = load_pipeline_model(args.checkpoint_dir, args.model, override_batch_size=args.batch_size, override_patch_size=args.patch_size)
     drab_volcano_legend = False
     if segmentation_model.name == 'drab volcano':
         log.warning('Drab Volcano uses a pretrained set of map units for segmentation and is not promptable by the legend')
@@ -453,7 +466,7 @@ def construct_amqp_pipeline(args):
     p = pipeline_manager(monitor=file_monitor())
     legend_model = yolo_legend_model()
     legend_model.load_model(args.checkpoint_dir)
-    segmentation_model = load_pipeline_model(args.checkpoint_dir, args.model, override_batch_size=args.batch_size)
+    segmentation_model = load_pipeline_model(args.checkpoint_dir, args.model, override_batch_size=args.batch_size, override_patch_size=args.patch_size)
     drab_volcano_legend = False
     if segmentation_model.name == 'drab volcano':
         log.warning('Drab Volcano uses a pretrained set of map units for segmentation and is not promptable by the legend')

requirements.txt

@@ -36,6 +36,7 @@ tqdm
 rich>13.0
 matplotlib==3.8.4
 
+
 ### Submodule Dependencies
 # Legend-extraction
 --extra-index-url https://pypi.nvidia.com
@@ -55,6 +56,14 @@ easyocr>=1.7.1
 pandas>=2.1.3
 joblib
 
+# Seasoned Lynx
+transformers
+peft
+pycocotools
+onnx
+onnxruntime
+fvcore
+
 ### OLD ####
 # mlflow
 #mlflow>=2.8.1

runTestPipeline.sh

@@ -0,0 +1,25 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_256"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+python pipeline.py \
+    -v --data ${SRC_DATA}/images \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 256

runTestPipeline_1024.sh

@@ -0,0 +1,29 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_1024_recolor"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+
+python pipeline.py \
+    -v --data ${SRC_DATA}/images \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 1024
+
+
+# make sure to change the patch size to the desired size

runTestPipeline_512.sh

@@ -0,0 +1,29 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_512"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+
+python pipeline.py \
+    -v --data ${SRC_DATA}/images \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 512
+
+
+# make sure to change the patch size to the desired size

runTestPipeline_512_recolor.sh

@@ -0,0 +1,29 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_512_update"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+
+python pipeline.py \
+    -v --data ${SRC_DATA}/images \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 512
+
+
+# make sure to change the patch size to the desired size

runTestPipeline_512_recolor_this.sh

@@ -0,0 +1,29 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_512_update_recolor"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+
+python pipeline.py \
+    -v --data ${SRC_DATA}/images \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 512
+
+
+# make sure to change the patch size to the desired size

runTestPipeline_debug.sh

@@ -0,0 +1,25 @@
+#TAG="0.4.2" #TAG="0.4.2"
+TAG="pr-26"
+IMAGE="criticalmaas-pipeline_${TAG}.sif"
+# apptainer pull ${IMAGE} docker://ncsa/criticalmaas-pipeline:${TAG}
+TEST_MODEL="seasoned_lynx"
+# DEST_DIR="/projects/bbym/shared/testingPR/${TAG}/${TEST_MODEL}/validation_w_layout"
+DEST_DIR="/projects/bcxi/shared/seasoned_lynx_testing/patch_1024_recolor"
+SRC_DATA="/projects/bcxi/shared/datasets/validation"
+mkdir -p ${DEST_DIR}/feedback
+mkdir -p ${DEST_DIR}/logs
+mkdir -p ${DEST_DIR}/output
+
+#srun -A bcxi-tgirails --time=4:00:00 --partition=gpu --ntasks-per-node=16 --gpus=4 --mem=50g --nodes=1 --pty /bin/bash
+
+# apptainer run --nv -B ${SRC_DATA}:/data -B ${DEST_DIR}/feedback:/feedback \
+#           -B ${DEST_DIR}/logs:/logs -B ${DEST_DIR}/output:/output ${IMAGE} \
+python pipeline.py -v \
+    --data /projects/bcxi/shared/datasets/validation/images/AR_Maumee.tif \
+	--legends ${SRC_DATA}/legends --output ${DEST_DIR}/output  \
+	--layouts ${SRC_DATA}/layouts \
+    --feedback ${DEST_DIR}/feedback \
+    --log ${DEST_DIR}/logs/log.log --model ${TEST_MODEL} \
+	--validation ${SRC_DATA}/true_segmentations \
+    --output_types raster_masks \
+    --patch_size 1024