Merge pull request #6 from jangsoopark/v2.0.0

V2.0.0

Author: jangsoopark

.gitignore

@@ -282,9 +282,14 @@ $RECYCLE.BIN/
 # asdf
 dataset/soc/raw/train/
 dataset/soc/raw/test/
-
 dataset/soc/train/
 dataset/soc/test/
+
+dataset/eoc-1/raw/train/
+dataset/eoc-1/raw/test/
+dataset/eoc-1/train/
+dataset/eoc-1/test/
+
 experiments/history
 
 *.pth

README.md

@@ -61,12 +61,14 @@ You can see the details in `src/model/_base.py` and `experiments/config/AConvNet
 ## Experiments
 
 You can download the MSTAR Dataset from [MSTAR Overview](https://www.sdms.afrl.af.mil/index.php?collection=mstar)
+- MSTAR Clutter - CD1 / CD2
+- MSTAR Target Chips (T72 BMP2 BTR70 SLICY) - CD1
+- MSTAR / IU Mixed Targets - CD1 / CD2
+- MSTAR / IU T-72 Variants - CD1 / CD2
+- MSTAR Predictlite Software - CD1
 
-### Standard Operating Condition (SOC)
 
-- MSTAR Target Chips (T72 BMP2 BTR70 SLICY) which is **MSTAR-PublicTargetChips-T72-BMP2-BTR70-SLICY.zip**
-- MSTAR / IU Mixed Targets which consists of **MSTAR-PublicMixedTargets-CD1.zip** and **MSTAR-PublicMixedTargets-CD2.zip**
-- **SLICY target is ignored**
+### Standard Operating Condition (SOC)
 
 |         |            | Train      |            | Test       |            |
 | ------- | ---------- | ---------- | ---------- | ---------- | ---------- |
@@ -82,7 +84,7 @@ You can download the MSTAR Dataset from [MSTAR Overview](https://www.sdms.afrl.a
 | ZIL-131 | E12        | 17         | 299        | 15         | 274        |
 | ZSU-234 | d08        | 17         | 299        | 15         | 274        |
 
-#### Training Set (Depression: 17$\degree$​)
+#### Training Set (Depression: 17$\degree$)
 
 ```shell
 MSTAR-PublicTargetChips-T72-BMP2-BTR70-SLICY
@@ -107,7 +109,7 @@ MSTAR-PublicMixedTargets-CD2/MSTAR_PUBLIC_MIXED_TARGETS_CD2
 
 ```
 
-#### Test Set (Depression: 15$\degree$​​)
+#### Test Set (Depression: 15$\degree$)
 
 ```shell
 MSTAR-PublicTargetChips-T72-BMP2-BTR70-SLICY
@@ -139,8 +141,8 @@ MSTAR-PublicMixedTargets-CD1/MSTAR_PUBLIC_MIXED_TARGETS_CD1
     - Place the two directories (`train` and  `test`) to the `dataset/raw`.
 ```shell
 $ cd src/data 
-$ python3 generate_dataset.py --is_train=True --use_phase=True
-$ python3 generate_dataset.py --is_train=False --use_phase=True
+$ python3 generate_dataset.py --is_train=True --use_phase=True --chip_size=94 --dataset=soc 
+$ python3 generate_dataset.py --is_train=False --use_phase=True --dataset=soc
 $ cd ..
 $ python3 train.py
 ```
@@ -169,6 +171,56 @@ $ python3 train.py
 
 ### Extended Operating Conditions (EOC)
 
+#### EOC-1 (Large depression angle change)
+
+
+|         | Train      |            |            | Test       |            |            |
+| ------- | ---------- | ---------- | ---------- | ---------- | ---------- | ---------- |
+| Class   | Serial No. | Depression | No. Images | Serial No. | Depression | No. Images |
+| T-72    | 132        | 17         | 232        | A64        | 30         | 196        |
+| 2S1     | b01        | 17         | 299        | b01        | 30         | 274        |
+| BRDM-2  | E-71       | 17         | 298        | E-71       | 30         | 274        |
+| ZSU-234 | d08        | 17         | 299        | d08        | 30         | 274        |
+
+```shell
+MSTAR-PublicTargetChips-T72-BMP2-BTR70-SLICY/MSTAR_PUBLIC_TARGETS_CHIPS_T72_BMP2_BTR70_SLICY
+├ TRAIN/17_DEG
+│    └ T72/SN_132/*.015   (232 images)
+└ ...
+
+MSTAR-PublicMixedTargets-CD2/MSTAR_PUBLIC_MIXED_TARGETS_CD2
+├ 17_DEG
+│    └ COL2/SCENE1
+│        ├ 2S1/*.000            (299 images)
+│        ├ BRDM_2/*.001         (298 images)
+│        └ ZSU_23_4/*.026       (299 images)
+└ ...
+
+```
+
+#### Test Set (Depression: 30$\degree$)
+
+```shell
+MSTAR-PublicT72Variants-CD2/MSTAR_PUBLIC_T_72_VARIANTS_CD2
+├ 30_DEG/COL2/SCENE1
+│    └ A64/*.024   (288 images)
+└ ...
+
+MSTAR-PublicMixedTargets-CD2/MSTAR_PUBLIC_MIXED_TARGETS_CD2
+├ 30_DEG
+│    └ COL2/SCENE1
+│        ├ 2S1/*.000            (288 images)
+│        ├ BRDM_2/*.001         (287 images)
+│        ├ ZSU_23_4/*.026       (288 images)
+│        └ ...
+└ ...
+
+```
+- Train Target: 2S1, BRDM2, T72, ZSU234 with depression angle 17$\degree$
+- Test Target: 2S1, BRDM2, T72, ZSU234 with depression angle 30$\degree$
+
+#### EOC-2 (Target configuration and version variants)
+
 ### Outlier Rejection
 
 ### End-to-End SAR-ATR Cases

dataset/eoc-1/raw/rename.py

@@ -0,0 +1,9 @@
+import glob
+import os
+
+
+target_list = glob.glob('./*/*')
+
+for name in target_list:
+    print(name, name.replace('_', ''))
+    os.rename(name, name.replace('_', ''))

experiments/config/AConvNet-EOC-1.json

@@ -0,0 +1,14 @@
+{
+  "model_name": "AConvNet-EOC-1",
+  "dataset": "eoc-1",
+  "num_classes": 4,
+  "channels": 2,
+  "batch_size": 100,
+  "epochs": 50,
+  "momentum": 0.9,
+  "lr": 1e-3,
+  "lr_step": [14],
+  "lr_decay": 0.1,
+  "weight_decay": 4e-3,
+  "dropout_rate": 0.5
+}

notebook/experiments-SOC.ipynb

@@ -145,7 +145,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "load test data set: 2425it [00:01, 2086.92it/s]\n",
+      "load test data set: 2425it [00:18, 134.26it/s]\n",
       "d:\\ivs\\project\\004-research\\signal-processing\\image-processing\\remote-sensing\\aconvnet\\aconvnet-pytorch\\venv\\lib\\site-packages\\torch\\nn\\functional.py:718: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at  ..\\c10/core/TensorImpl.h:1156.)\n",
       "  return torch.max_pool2d(input, kernel_size, stride, padding, dilation, ceil_mode)\n"
      ]
@@ -288,7 +288,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 105,
+   "execution_count": 14,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -305,7 +305,7 @@
     "        for _ in range(n)\n",
     "    ])\n",
     "    noise = noise * portions\n",
-    "    \n",
+    "        \n",
     "    return _images + noise.astype(np.float32)\n",
     "\n",
     "\n",
@@ -335,17 +335,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 106,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "ratio = 0.01, accuracy = 98.23\n",
-      "ratio = 0.05, accuracy = 94.19\n",
-      "ratio = 0.10, accuracy = 85.28\n",
-      "ratio = 0.15, accuracy = 73.40\n"
+      "ratio = 0.01, accuracy = 98.52\n",
+      "ratio = 0.05, accuracy = 94.56\n",
+      "ratio = 0.10, accuracy = 85.11\n",
+      "ratio = 0.15, accuracy = 73.11\n"
      ]
     }
    ],

src/data/generate_dataset.py

@@ -15,17 +15,22 @@
 flags.DEFINE_string('dataset', default='soc', help='')
 flags.DEFINE_boolean('is_train', default=True, help='')
 flags.DEFINE_boolean('use_phase', default=False, help='')
+flags.DEFINE_integer('chip_size', default=94, help='')
 FLAGS = flags.FLAGS
 
 project_root = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
 
-def generate(src_path, dst_path, is_train, use_phase):
+def generate(src_path, dst_path, is_train, use_phase, chip_size, dataset):
+    if not os.path.exists(src_path):
+        return
     if not os.path.exists(dst_path):
         os.makedirs(dst_path, exist_ok=True)
     print(f'Target Name: {os.path.basename(dst_path)}')
 
-    _mstar = mstar.MSTAR(is_train=is_train, use_phase=use_phase, patch_size=88, stride=1)
+    _mstar = mstar.MSTAR(
+        name=dataset, is_train=is_train, use_phase=use_phase, chip_size=chip_size, patch_size=88, stride=1
+    )
 
     image_list = glob.glob(os.path.join(src_path, '*'))
 
@@ -52,8 +57,8 @@ def main(_):
         (
             os.path.join(raw_root, mode, target),
             os.path.join(output_root, target),
-            FLAGS.is_train, FLAGS.use_phase
-        ) for target in mstar.target_name
+            FLAGS.is_train, FLAGS.use_phase, FLAGS.chip_size, FLAGS.dataset
+        ) for target in mstar.target_name_soc
     ]
 
     with Pool(10) as p:

src/data/mstar.py

@@ -6,15 +6,32 @@
 import glob
 import os
 
-target_name = ('2S1', 'BMP2', 'BRDM2', 'BTR60', 'BTR70', 'D7', 'T62', 'T72', 'ZIL131', 'ZSU234')
-serial_number = ('b01', '9563', 'E-71', 'k10yt7532', 'c71', '92v13015', 'A51', '132', 'E12', 'd08')
+target_name_soc = ('2S1', 'BMP2', 'BRDM2', 'BTR60', 'BTR70', 'D7', 'T62', 'T72', 'ZIL131', 'ZSU234')
+target_name_eoc_1 = ('2S1', 'BRDM2', 'T72', 'ZSU234')
+serial_number = {
+    'b01': 0,
+    '9563': 1,
+    'E-71': 2,
+    'k10yt7532': 3,
+    'c71': 4,
+    '92v13015': 5,
+    'A51': 6,
+
+    '132': 7,
+    'A64': 7,
+
+    'E12': 8,
+    'd08': 9
+}
 
 
 class MSTAR(object):
 
-    def __init__(self, is_train=False, use_phase=False, patch_size=88, stride=40):
+    def __init__(self, name='soc', is_train=False, use_phase=False, chip_size=94, patch_size=88, stride=40):
+        self.name = name
         self.is_train = is_train
         self.use_phase = use_phase
+        self.chip_size = chip_size
         self.patch_size = patch_size
         self.stride = stride
 
@@ -74,22 +91,22 @@ def _center_crop(data, size=128):
 
         return data[y: y + size, x: x + size]
 
-    @staticmethod
-    def _data_augmentation(data, patch_size=88, stride=40):
+    def _data_augmentation(self, data, patch_size=88, stride=40):
         # patch extraction
-        _data = MSTAR._center_crop(data, size=94)
+        _data = MSTAR._center_crop(data, size=self.chip_size)
         _, _, channels = _data.shape
         patches = shape.view_as_windows(_data, window_shape=(patch_size, patch_size, channels), step=stride)
         patches = patches.reshape(-1, patch_size, patch_size, channels)
         return patches
 
-    @staticmethod
-    def _extract_meta_label(header):
+    def _extract_meta_label(self, header):
 
         target_type = header['TargetType']
         sn = header['TargetSerNum']
 
-        class_id = serial_number.index(sn)
+        class_id = serial_number[sn]
+        if self.name == 'eoc-1':
+            class_id = target_name_eoc_1.index(target_name_soc[class_id])
 
         azimuth_angle = MSTAR._get_azimuth_angle(header['TargetAz'])
 

src/train.py

@@ -22,7 +22,7 @@
 flags.DEFINE_string('config_name', 'config/AConvNet-SOC.json', help='')
 FLAGS = flags.FLAGS
 
-common.set_random_seed(12321)
+#common.set_random_seed(12321)
 
 
 def load_dataset(path, is_train, name, batch_size):