-
Notifications
You must be signed in to change notification settings - Fork 0
/
tutorial6_predict_KHQ.sh
435 lines (375 loc) · 15.3 KB
/
tutorial6_predict_KHQ.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
#!/bin/bash
__doc__="
This tutorial is a predict-time demonstration on KHQ similar to a system that
would be run in production.
"
### DEFINE VARIABLES
# TODO: add instructions for how to set these if they are unset.
#
# If you can access our DVC repo:
DVC_DATA_DPATH=$(geowatch_dvc --tags='phase2_data' --hardware=ssd)
# The "name" of the demo dataset we will create
DATASET_SUFFIX=KHQ_Tutorial6_Data
# Set this to where you want to build the dataset
DEMO_DPATH=$DVC_DATA_DPATH/$DATASET_SUFFIX
REGION_ID=KHQ_R001
RAW_DSET_DPATH=$DEMO_DPATH/Aligned-$DATASET_SUFFIX
TIMECOMBO_DSET_DPATH=$DEMO_DPATH/TimeCombine-$DATASET_SUFFIX
# ==================================
# Create a demo region file, and create vairables that point at relevant
# paths, which are by default written in your ~/.cache folder
python -m geowatch.demo.demo_region
REGION_FPATH="$HOME/.cache/geowatch/demo/annotations/${REGION_ID}.geojson"
SITE_GLOBSTR="$HOME/.cache/geowatch/demo/annotations/${REGION_ID}_sites/*.geojson"
mkdir -p "$DEMO_DPATH"
# This is a string code indicating what STAC endpoint we will pull from
SENSORS="sentinel-2-l2a,landsat-c2l2-sr,landsat-c2l2-bt"
# Depending on the STAC endpoint, some parameters may need to change:
# collated - True for IARPA endpoints, Usually False for public data
# requester_pays - True for public landsat
# api_key - A secret for non public data
export REQUESTER_PAYS=True
export SMART_STAC_API_KEY=""
export GDAL_DISABLE_READDIR_ON_OPEN=EMPTY_DIR
# Construct the TA2-ready dataset.
# This is a cmdqueue pipeline of simpler commands
python -m geowatch.cli.queue_cli.prepare_ta2_dataset \
--dataset_suffix=$DATASET_SUFFIX \
--cloud_cover=30 \
--stac_query_mode=auto \
--sensors "$SENSORS" \
--api_key=env:SMART_STAC_API_KEY \
--collated False \
--requester_pays=$REQUESTER_PAYS \
--dvc_dpath="$DEMO_DPATH" \
--aws_profile=iarpa \
--region_globstr="$REGION_FPATH" \
--site_globstr="$SITE_GLOBSTR" \
--fields_workers=8 \
--convert_workers=0 \
--align_workers=26 \
--cache=1 \
--skip_existing=0 \
--ignore_duplicates=1 \
--target_gsd="10GSD" \
--visualize=False \
--max_products_per_region=100 \
--backend=serial \
--run=0
# Create a low temporal resolution time-combined dataset
# (We will use this for BAS)
python -m geowatch.cli.coco_time_combine \
--kwcoco_fpath="$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--output_kwcoco_fpath="$TIMECOMBO_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--channels="red|green|blue|nir|swir16|swir22|pan|coastal|cirrus|B05|B06|B07|B8A|B09" \
--resolution="10GSD" \
--time_window=1y \
--remove_seasons=winter \
--merge_method=median \
--spatial_tile_size=1024 \
--mask_low_quality=True \
--start_time=2010-03-01 \
--assets_dname="raw_bands" \
--workers=0
# COLD FEATURES on original data
#
# Helper to generate the raw commands
# python -m geowatch.cli.queue_cli.prepare_teamfeats --src_kwcocos "$IMGONLY_COCO_FPATH" --expt_dvc_dpath="$DVC_EXPT_DPATH" \
# --with_cold=1 --skip_existing=1 --gres=0,1 --tmux_workers=4 \
# --backend=tmux --run=0 --print-commands
python -m geowatch.tasks.cold.predict \
--coco_fpath="$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--mod_coco_fpath="$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}_cold.kwcoco.zip" \
--out_dpath="$RAW_DSET_DPATH/${REGION_ID}" \
--workermode=process \
--workers=0 \
--sensors=L8,S2 \
--adj_cloud=False \
--method=COLD \
--prob=0.99 \
--conse=8 \
--cm_interval=60 \
--year_lowbound=None \
--year_highbound=None \
--coefs=cv,rmse,a0,a1,b1,c1 \
--coefs_bands=0,1,2,3,4,5 \
--timestamp=False \
--combine=False \
--resolution=10GSD
geowatch visualize "$DEMO_DPATH"/Aligned-KHQ_Tutorial6_Data/KHQ_R001/imgonly-KHQ_R001_cold.kwcoco.zip --smart=1 \
--channels="(L8,S2):(red|green|blue,red_COLD_a1|green_COLD_a1|blue_COLD_a1,red_COLD_cv|green_COLD_cv|blue_COLD_cv,red_COLD_rmse|green_COLD_rmse|blue_COLD_rmse)"
python -m geowatch.cli.coco_combine_features \
--src \
"$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
"$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}_cold.kwcoco.zip" \
--dst="$RAW_DSET_DPATH/${REGION_ID}/combo_imgonly-${REGION_ID}_C.kwcoco.zip"
python -m geowatch.tasks.cold.transfer_features \
--coco_fpath "$RAW_DSET_DPATH/${REGION_ID}/combo_imgonly-${REGION_ID}_C.kwcoco.zip" \
--combine_fpath "$TIMECOMBO_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--new_coco_fpath "$TIMECOMBO_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-C.kwcoco.zip" \
--copy_assets=0 \
--io_workers=4
geowatch visualize \
/home/joncrall/remote/toothbrush/data/dvc-repos/smart_data_dvc-ssd/KHQ_Tutorial6_Data/TimeCombine-KHQ_Tutorial6_Data/KHQ_R001/imgonly-KHQ_R001-C.kwcoco.zip \
--smart=1 --channels="(L8,S2):(red|green|blue,red_COLD_a1|green_COLD_a1|blue_COLD_a1,red_COLD_cv|green_COLD_cv|blue_COLD_cv,red_COLD_rmse|green_COLD_rmse|blue_COLD_rmse)"
inspect_stuff(){
IMGONLY_COCO_FPATH="$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip"
geowatch stats "$IMGONLY_COCO_FPATH"
geowatch visualize "$IMGONLY_COCO_FPATH" --smart
geowatch model_stats "$BAS_MODEL_FPATH"
}
DVC_DATA_DPATH=$(geowatch_dvc --tags='phase2_data' --hardware=ssd)
DATASET_SUFFIX=KHQ_Tutorial6_Data
DEMO_DPATH=$DVC_DATA_DPATH/$DATASET_SUFFIX
IMGONLY_COCO_FPATH="$DEMO_DPATH/Aligned-$DATASET_SUFFIX/KHQ_R001/imgonly-KHQ_R001-rawbands.kwcoco.zip"
DVC_EXPT_DPATH=$(geowatch_dvc --tags='phase2_expt' --hardware=auto)
BAS_MODEL_FPATH=$DVC_EXPT_DPATH/models/fusion/Drop7-MedianNoWinter10GSD/packages/Drop7-MedianNoWinter10GSD_bgrn_split6_V74/Drop7-MedianNoWinter10GSD_bgrn_split6_V74_epoch46_step4042.pt
ACSC_MODEL_FPATH=$DVC_EXPT_DPATH/models/fusion/Drop4-SC/packages/Drop4_tune_V30_8GSD_V3/Drop4_tune_V30_8GSD_V3_epoch=2-step=17334.pt.pt
PIPELINE_OUTPUT_DPATH="$DEMO_DPATH/outputs"
echo "BAS_MODEL_FPATH = $BAS_MODEL_FPATH"
echo "ACSC_MODEL_FPATH = $ACSC_MODEL_FPATH"
test -f "$BAS_MODEL_FPATH" || echo "missing BAS model"
test -f "$ACSC_MODEL_FPATH" || echo "missing ACSC model"
python -m geowatch.tasks.fusion.predict \
--package_fpath="$BAS_MODEL_FPATH" \
--test_dataset="$TIMECOMBO_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--pred_dataset="$PIPELINE_OUTPUT_DPATH"/bas_pxl/pred.kwcoco.zip \
--chip_overlap=0.3 \
--chip_dims "196,196" \
--time_span=auto \
--fixed_resolution=10GSD \
--time_sampling=soft4 \
--drop_unused_frames=True \
--with_saliency=True \
--with_class=False \
--with_change=False \
--num_workers=2 \
--devices=0, \
--batch_size=1
python -m geowatch.cli.run_tracker \
--input_kwcoco "$PIPELINE_OUTPUT_DPATH"/bas_pxl/pred.kwcoco.zip \
--default_track_fn saliency_heatmaps \
--track_kwargs '{
"agg_fn": "probs",
"thresh": 0.37,
"time_thresh": 0.8,
"inner_window_size": "1y",
"inner_agg_fn": "max",
"norm_ord": "inf",
"moving_window_size": null,
"poly_merge_method": "v2",
"polygon_simplify_tolerance": 1,
"min_area_square_meters": 7200,
"max_area_square_meters": 8000000
}' \
--clear_annots=True \
--out_site_summaries_fpath "$PIPELINE_OUTPUT_DPATH/bas_poly/site_summaries_manifest.json" \
--out_site_summaries_dir "$PIPELINE_OUTPUT_DPATH/bas_poly/site_summaries" \
--out_sites_fpath "$PIPELINE_OUTPUT_DPATH/bas_poly/sites_manifest.json" \
--out_sites_dir "$PIPELINE_OUTPUT_DPATH/bas_poly/sites" \
--out_kwcoco "$PIPELINE_OUTPUT_DPATH/bas_poly/poly.kwcoco.zip" \
--viz_out_dir "$PIPELINE_OUTPUT_DPATH/bas_poly/_viz_tracker" \
--site_summary=None
python -m geowatch.cli.cluster_sites \
--context_factor=1.0 \
--minimum_size=128x128@2GSD \
--maximum_size=256x256@2GSD \
--src="$PIPELINE_OUTPUT_DPATH/bas_poly/site_summaries_manifest.json" \
--dst_dpath="$PIPELINE_OUTPUT_DPATH/ac_clusters" \
--dst_region_fpath="$PIPELINE_OUTPUT_DPATH/ac_clusters/clustered.geojson" \
--io_workers=4 \
--draw_clusters=True \
--crop_time=True
python -m geowatch.cli.coco_align \
--src "$RAW_DSET_DPATH/${REGION_ID}/imgonly-${REGION_ID}-rawbands.kwcoco.zip" \
--dst "$PIPELINE_OUTPUT_DPATH/ac_crops/sitecrop.kwcoco.zip" \
--regions="$PIPELINE_OUTPUT_DPATH/ac_clusters/clustered.geojson" \
--site_summary=True \
--img_workers=16 \
--aux_workers=2 \
--verbose=1 \
--debug_valid_regions=False \
--visualize=False \
--keep=img \
--geo_preprop=auto \
--force_nodata=-9999 \
--include_channels='red|green|blue|nir|quality' \
--exclude_sensors=L8 \
--minimum_size=128x128@2GSD \
--convexify_regions=True \
--target_gsd=2 \
--context_factor=1.5 \
--force_min_gsd=8 \
--rpc_align_method=orthorectify
python -m geowatch.tasks.fusion.predict \
--package_fpath="$ACSC_MODEL_FPATH" \
--test_dataset="$PIPELINE_OUTPUT_DPATH/ac_crops/sitecrop.kwcoco.zip" \
--pred_dataset="$PIPELINE_OUTPUT_DPATH/ac_pxl/ac_heatmaps.kwcoco.zip" \
--tta_fliprot=0.0 \
--tta_time=0.0 \
--chip_overlap=0.3 \
--fixed_resolution=8GSD \
--output_space_scale=8GSD \
--time_span=6m \
--time_sampling=auto \
--time_steps=12 \
--chip_dims=auto \
--set_cover_algo=None \
--resample_invalid_frames=3 \
--observable_threshold=0.0 \
--mask_low_quality=True \
--drop_unused_frames=True \
--write_workers=0 \
--with_saliency=True \
--with_class=True \
--with_change=False \
--saliency_chan_code=ac_salient \
--num_workers=12 \
--batch_size=1 \
--devices=0,
python -m geowatch.cli.run_tracker \
--input_kwcoco "$PIPELINE_OUTPUT_DPATH/ac_pxl/ac_heatmaps.kwcoco.zip" \
--default_track_fn class_heatmaps \
--track_kwargs '{"boundaries_as": "polys", "thresh": 0.07, "resolution": "8GSD", "min_area_square_meters": 7200}' \
--clear_annots=True \
--out_site_summaries_fpath "$PIPELINE_OUTPUT_DPATH/ac_poly/site_summaries_manifest.json" \
--out_site_summaries_dir "$PIPELINE_OUTPUT_DPATH/ac_poly/site_summaries" \
--out_sites_fpath "$PIPELINE_OUTPUT_DPATH/ac_poly/sites_manifest.json" \
--out_sites_dir "$PIPELINE_OUTPUT_DPATH/ac_poly/sites" \
--out_kwcoco "$PIPELINE_OUTPUT_DPATH/ac_poly/poly.kwcoco.zip" \
--site_summary "$PIPELINE_OUTPUT_DPATH/bas_poly/site_summaries_manifest.json" \
--boundary_region=None
# A basic pipeline can be run as a schedule evaluation pipeline.
python -m geowatch.mlops.schedule_evaluation --params="
pipeline: full
matrix:
######################
## BAS PIXEL PARAMS ##
######################
bas_pxl.package_fpath:
- $BAS_MODEL_FPATH
bas_pxl.test_dataset:
- $IMGONLY_COCO_FPATH
bas_pxl.chip_overlap: 0.3
bas_pxl.chip_dims:
- [196,196]
bas_pxl.time_span: auto
bas_pxl.fixed_resolution: 10GSD
bas_pxl.time_sampling: soft4
######################
## BAS POLY PARAMS ##
######################
bas_poly.thresh: 0.425
bas_poly.time_thresh: 0.8
bas_poly.inner_window_size: 1y
bas_poly.inner_agg_fn: max
bas_poly.norm_ord: inf
bas_poly.moving_window_size: null
bas_poly.poly_merge_method: 'v2'
bas_poly.polygon_simplify_tolerance: 1
bas_poly.agg_fn: probs
bas_poly.min_area_square_meters: 7200
bas_poly.max_area_square_meters: 8000000
###########################
## BAS POLY EVAL PARAMS ##
###########################
bas_poly_eval.true_site_dpath: $LORES_DVC_DATA_DPATH/annotations/drop6/site_models
bas_poly_eval.true_region_dpath: $LORES_DVC_DATA_DPATH/annotations/drop6/region_models
### SV
sv_crop.enabled: 1
sv_crop.minimum_size: '256x256@2GSD'
sv_crop.num_start_frames: 3
sv_crop.num_end_frames: 3
sv_crop.context_factor: 1.6
sv_dino_boxes.enabled: 1
sv_dino_boxes.package_fpath: $DVC_EXPT_DPATH/models/kitware/xview_dino.pt
sv_dino_boxes.window_dims: 256
sv_dino_boxes.window_overlap: 0.5
sv_dino_boxes.fixed_resolution: 3GSD
sv_dino_filter.enabled: 1
sv_dino_filter.end_min_score:
- 0.15
sv_dino_filter.start_max_score: 1.0
sv_dino_filter.box_score_threshold: 0.01
sv_dino_filter.box_isect_threshold: 0.1
sv_depth_score.enabled: 1
sv_depth_score.model_fpath:
- $DVC_EXPT_DPATH/models/depth_pcd/basicModel2.h5
sv_depth_filter.threshold:
- 0.10
##########################
## Cluster Sites Params ##
##########################
cluster_sites.context_factor: 1.5
cluster_sites.minimum_size: '128x128@8GSD'
cluster_sites.maximum_size: '1024x1024@8GSD'
########################
## SC CROPPING PARAMS ##
########################
sc_crop.force_nodata: -9999
sc_crop.include_channels: 'red|green|blue|nir|quality'
sc_crop.exclude_sensors: 'L8'
sc_crop.minimum_size: '128x128@8GSD'
sc_crop.convexify_regions: True
sc_crop.target_gsd: 2
sc_crop.context_factor: 1.5
sc_crop.force_min_gsd: 8
sc_crop.img_workers: 16
sc_crop.aux_workers: 2
#####################
## SC PIXEL PARAMS ##
#####################
sc_pxl.package_fpath:
- $ACSC_MODEL_FPATH
sc_pxl.tta_fliprot: 0.0
sc_pxl.tta_time: 0.0
sc_pxl.chip_overlap: 0.3
sc_pxl.fixed_resolution: 8GSD
sc_pxl.output_space_scale: 8GSD
sc_pxl.time_span: 6m
sc_pxl.time_sampling: auto
sc_pxl.time_steps: 12
sc_pxl.chip_dims: auto
sc_pxl.set_cover_algo: null
sc_pxl.resample_invalid_frames: 3
sc_pxl.observable_threshold: 0.0
sc_pxl.mask_low_quality: true
sc_pxl.drop_unused_frames: true
sc_pxl.num_workers: 12
sc_pxl.batch_size: 1
sc_pxl.write_workers: 0
#####################
## SC POLY PARAMS ##
#####################
sc_poly.thresh: 0.07
sc_poly.boundaries_as: polys
sc_poly.resolution: 8GSD
sc_poly.min_area_square_meters: 7200
##########################
## SC POLY EVAL PARAMS ##
##########################
sc_poly_eval.true_site_dpath: $LORES_DVC_DATA_DPATH/annotations/drop6/site_models
sc_poly_eval.true_region_dpath: $LORES_DVC_DATA_DPATH/annotations/drop6/region_models
##################################
## HIGH LEVEL PIPELINE CONTROLS ##
##################################
bas_pxl.enabled: 1
bas_poly.enabled: 1
sc_crop.enabled: 1
sc_pxl.enabled: 1
sc_poly.enabled: 1
sc_poly_eval.enabled: 0
bas_pxl_eval.enabled: 0
bas_poly_eval.enabled: 0
sc_pxl_eval.enabled: 0
bas_poly_viz.enabled: 0
sc_poly_viz.enabled: 0
submatrices1:
- bas_pxl.test_dataset: $IMGONLY_COCO_FPATH
sc_crop.crop_src_fpath: $IMGONLY_COCO_FPATH
" \
--root_dpath="$DVC_EXPT_DPATH/_demo_khq" \
--queue_name "_demo_khq" \
--devices="0,1" \
--backend=tmux --tmux_workers=6 \
--cache=1 --skip_existing=0 --run=0