# evaluateSaliency

**Repository Path**: ericps/evaluateSal

## Basic Information

- **Project Name**: evaluateSaliency
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-03-29
- **Last Updated**: 2022-01-14

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

### attention dataset
```bash
Conan1 CC=0.4178, NSS=9.0314, KL=2.1233, AUC=0.9956, sAUC=0.9936
Skiing CC=0.3735, NSS=7.8986, KL=2.3488, AUC=0.9939, sAUC=0.9768
Alien CC=0.3782, NSS=8.224, KL=2.3367, AUC=0.9941, sAUC=0.9794
Conan2 CC=0.3809, NSS=8.6427, KL=2.2883, AUC=0.9951, sAUC=0.9952
Surfing CC=0.3185, NSS=7.4912, KL=2.7425, AUC=0.9932, sAUC=0.9665
War CC=0.3331, NSS=8.1059, KL=2.6365, AUC=0.9943, sAUC=0.97
Cooking CC=0.3579, NSS=7.2651, KL=2.4079, AUC=0.9926, sAUC=0.9832
Football CC=0.3626, NSS=8.7277, KL=2.4525, AUC=0.9951, sAUC=0.9845
Rhinos CC=0.3662, NSS=8.5873, KL=2.4005, AUC=0.9948, sAUC=0.9703
```

### sample attention dataset
```bash
CC=0.3587, NSS=8.1267, KL=2.4531, AUC=0.9941 sAUC=0.9502
```

attention论文中的结果：

<img src="https://s2.loli.net/2022/01/14/v9u2DySt87xK6kZ.png" style="zoom:80%;" />  



### deepCNN

#### test 
- **shuffle**: `CC=0.1512, NSS=3.4033, KL=4.9944, AUC=0.7849 sAUC=0.7008`
- `CC=0.2047, NSS=6.6387, KL=3.6678, AUC=0.9814 sAUC=0.7874`
- **no shuffle**: `CC=0.0147, NSS=0.4306, KL=6.3465, AUC=0.6506 sAUC=0.4996`
- `CC=0.0273, NSS=0.7747, KL=6.1994, AUC=0.7232 sAUC=0.7553`

#### train
##### 归一化
- `CC=0.1102, NSS=2.4659, KL=5.0099, AUC=0.8675 sAUC=0.8548`
- `CC=0.1799, NSS=4.1379, KL=4.0874, AUC=0.9185 sAUC=0.8966`
- `CC=0.1474, NSS=3.4652, KL=4.6141, AUC=0.8556 sAUC=0.8001`
- `CC=0.1544, NSS=3.6161, KL=4.4892, AUC=0.8747 sAUC=0.7208`

##### 没有归一化
- **model1**: `CC=0.1102, NSS=2.4659, KL=5.0099, AUC=0.8675`
- **model2**: `CC=0.1799, NSS=4.1379, KL=4.0874, AUC=0.9185`
- **model3**: `CC=0.1474, NSS=3.4652, KL=4.6141, AUC=0.8556`
- **model4**: `CC=0.1544, NSS=3.6161, KL=4.4890, AUC=0.8747 sAUC=0.7208`

### s2cnn
#### test
- **shuffle**:`CC=0.1683, NSS=4.5349, KL=3.941, AUC=0.9701 sAUC=0.835`

#### train
- **sphere_kl_relu.pth**: `CC=0.1496, NSS=3.4171 KL=4.3519, AUC=0.9212 sAUC=0.8053`