Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Commit fa5b770

Browse files
swuxyjswuxyj
committed
add cifar10 config
1 parent 15429eb commit fa5b770

12 files changed

Lines changed: 141 additions & 60037 deletions

File tree

‎DFH.py‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -31,8 +31,10 @@ def get_config():
3131
"net": AlexNet,
3232
# "net":ResNet,
3333
# "dataset": "cifar10",
34+
"dataset": "cifar10-1",
35+
# "dataset": "cifar10-2",
3436
# "dataset": "coco",
35-
"dataset": "mirflickr",
37+
# "dataset": "mirflickr",
3638
# "dataset": "voc2012
3739
# "dataset": "nuswide_21",
3840
# "dataset": "nuswide_21_m",

‎DHN.py‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -25,10 +25,10 @@ def get_config():
2525
"batch_size": 128,
2626
"net": AlexNet,
2727
# "net":ResNet,
28-
# "dataset": "cifar10",
28+
"dataset": "cifar10",
2929
# "dataset": "mirflickr",
3030
# "dataset": "voc2012
31-
"dataset": "nuswide_21",
31+
# "dataset": "nuswide_21",
3232
# "dataset": "nuswide_21_m",
3333
# "dataset": "nuswide_81_m",
3434
# "dataset": "coco",

‎DPSH.py‎

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -18,17 +18,17 @@
1818
def get_config():
1919
config = {
2020
"alpha": 0.1,
21-
# "optimizer":{"type": optim.SGD, "optim_params": {"lr": 0.05, "weight_decay": 10 ** -5}, "lr_type": "step"},
21+
# "optimizer": {"type": optim.SGD, "optim_params": {"lr": 0.005, "weight_decay": 10 ** -5}, "lr_type": "step"},
2222
"optimizer": {"type": optim.RMSprop, "optim_params": {"lr": 1e-5, "weight_decay": 10 ** -5}, "lr_type": "step"},
2323
"info": "[DPSH]",
2424
"resize_size": 256,
2525
"crop_size": 224,
26-
"batch_size": 64,
26+
"batch_size": 128,
2727
"net": AlexNet,
2828
# "net":ResNet,
29-
# "dataset": "cifar10",
29+
"dataset": "cifar13",
3030
# "dataset": "coco",
31-
"dataset": "mirflickr",
31+
# "dataset": "mirflickr",
3232
# "dataset": "voc2012",
3333
# "dataset":"imagenet",
3434
# "dataset": "nuswide_21",
@@ -62,7 +62,7 @@ def forward(self, u, y, ind, config):
6262
s = (y @ self.Y.t() > 0).float()
6363
inner_product = u @ self.U.t() * 0.5
6464

65-
likelihood_loss = (1 + (-inner_product.abs()).exp()).log() + inner_product.clamp(min=0) - s * inner_product
65+
likelihood_loss = (1 + (-(inner_product.abs())).exp()).log() + inner_product.clamp(min=0) - s * inner_product
6666

6767
likelihood_loss = likelihood_loss.mean()
6868

‎DSDH.py‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -31,7 +31,9 @@ def get_config():
3131
"net": AlexNet,
3232
# "net":ResNet,
3333
# "dataset": "cifar10",
34-
"dataset": "mirflickr",
34+
# "dataset": "cifar10-1",
35+
"dataset": "cifar10-2",
36+
# "dataset": "mirflickr",
3537
# "dataset": "voc2012
3638
# "dataset": "nuswide_21",
3739
# "dataset": "nuswide_21_m",

‎DSH.py‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -26,9 +26,9 @@ def get_config():
2626
"net": AlexNet,
2727
# "net":ResNet,
2828
# "dataset": "cifar10",
29-
"dataset": "mirflickr",
30-
# "dataset": "voc2012
31-
# "dataset": "nuswide_21",
29+
# "dataset": "mirflickr",
30+
# "dataset": "voc2012",
31+
"dataset": "nuswide_21",
3232
# "dataset": "nuswide_21_m",
3333
# "dataset": "nuswide_81_m",
3434
# "dataset": "coco",

‎DTSH.py‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -27,10 +27,10 @@ def get_config():
2727
"batch_size": 128,
2828
"net": AlexNet,
2929
# "net":ResNet,
30-
# "dataset": "cifar10",
30+
"dataset": "cifar13",
3131
# "dataset": "coco",
3232
# "dataset":"imagenet",
33-
"dataset": "mirflickr",
33+
# "dataset": "mirflickr",
3434
# "dataset": "voc2012
3535
# "dataset": "nuswide_21",
3636
# "dataset": "nuswide_21_m",

‎GreedyHash.py‎

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -28,13 +28,14 @@ def get_config():
2828
"batch_size": 64,
2929
"net": AlexNet,
3030
# "net":ResNet,
31-
"dataset": "cifar10",
31+
# "dataset": "cifar10",
32+
# "dataset": "cifar10-1",
33+
"dataset": "cifar10-2",
3234
# "dataset": "coco",
3335
# "dataset": "nuswide_21",
3436
# "dataset": "imagenet",
3537
"epoch": 200,
3638
"test_map": 15,
37-
"save_path": "save/GreedyHash",
3839
"GPU": True,
3940
# "GPU":False,
4041
"bit_list": [48],
@@ -149,4 +150,4 @@ def train_val(config, bit):
149150
config = get_config()
150151
print(config)
151152
for bit in config["bit_list"]:
152-
train_val(config, bit)
153+
train_val(config, bit)

‎README.md‎

Lines changed: 11 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -18,10 +18,13 @@ pyhon DSDH.py
1818
```
1919

2020
# Dataset
21+
There are three different configurations for cifar10
22+
- config["dataset"]="cifar10" will use 1000 images (100 images per class) as the query set, 5000 images( 500 images per class) as training set , the remaining 54,000 images are used as database.
23+
- config["dataset"]="cifar10-1" will use 1000 images (100 images per class) as the query set, the remaining 59,000 images are used as database, 5000 images( 500 images per class) are randomly sampled from the database as training set.
24+
- config["dataset"]="cifar10-2" will use 10000 images (1000 images per class) as the query set, 50000 images( 5000 images per class) as training set , the remaining 54,000 images are used as database and training set.
25+
2126
You can download ImageNet, NUS-WIDE-m and COCO dataset [here](https://github.com/thuml/HashNet/tree/master/pytorch)
2227

23-
download cifar10(Lossless PNG format) [here](https://drive.google.com/open?id=1NZ5QKW2zqzN-RQ4VDpuOAb-UgcsTPUJK)
24-
2528
and download NUS-WIDE [here](https://github.com/TreezzZ/DSDH_PyTorch)
2629

2730
NUS-WIDE-m is different from NUS-WIDE, so i made a distinction.
@@ -145,7 +148,7 @@ code [IDHN-Tensorflow](https://github.com/pectinid16/IDHN)
145148
<td ></td><td >mirflickr</td> <td >0.775</td> <td >-</td>
146149
</tr>
147150
<tr>
148-
<td >DSDH</td><td >cifar10</td> <td >0.755</td> <td >0.820</td>
151+
<td >DSDH</td><td >cifar10-1</td> <td >0.755</td> <td >0.820</td>
149152
</tr>
150153
<tr>
151154
<td ></td><td >nus_wide_21</td> <td >0.819</td> <td >0.829</td>
@@ -175,7 +178,7 @@ code [IDHN-Tensorflow](https://github.com/pectinid16/IDHN)
175178
<td ></td><td >mirflickr</td> <td >0.753</td> <td >-</td>
176179
</tr>
177180
<tr>
178-
<td >DFH</td><td >cifar 10</td> <td >0.783</td> <td >0.844</td>
181+
<td >DFH</td><td >cifar10-1</td> <td >0.785</td> <td >0.844</td>
179182
</tr>
180183
<tr>
181184
<td ></td><td >nus_wide_21</td> <td >0.834</td> <td >0.842</td>
@@ -190,7 +193,10 @@ code [IDHN-Tensorflow](https://github.com/pectinid16/IDHN)
190193
<td ></td><td >mirflickr</td> <td >0.766</td> <td >-</td>
191194
</tr>
192195
<tr>
193-
<td >GreedyHash</td><td >cifar 10</td> <td >0.798</td> <td >0.822</td>
196+
<td >GreedyHash</td><td >cifar10-1</td> <td >0.796</td> <td >0.822</td>
197+
</tr>
198+
<tr>
199+
<td ></td><td >cifar10-2</td> <td >0.932</td> <td >0.944</td>
194200
</tr>
195201
<tr>
196202
<td ></td><td >imagenet</td> <td >0.678</td> <td >0.688</td>

0 commit comments

Comments
 (0)