Add version in requirements.txt

README.md

@@ -13,7 +13,9 @@ This offical implementation of PKD (Pseudo Knowledge Distillation) from On-devic
 **This code is based on the implementation of  [EML-NET-Saliency](https://github.com/SenJia/EML-NET-Saliency), [SimpleNet](https://github.com/samyak0210/saliency), [MSI-Net](https://github.com/alexanderkroner/saliency), and [EEEA-Net](https://github.com/chakkritte/EEEA-Net).**
 
 ## Prerequisite for server
- - Tested on Ubuntu OS version 20.04.x
+ - Tested on Ubuntu OS version 20.04.4 LTS
+ - Tested on Python 3.6.13
+ - Tested on CUDA 11.6
  - Tested on PyTorch 1.10.2 and TorchVision 0.11.3
  - Tested on NVIDIA V100 32 GB (four cards)
 
@@ -55,15 +57,15 @@ PKD
 ### Creating new environments
 
 ```
-conda create -n pkd python=3.6
+conda create -n pkd python=3.6.13
 conda activate pkd
 conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch
 ```
 
 ### Install Requirements
 
 ```
-pip install -r requirements.txt
+pip install -r requirements.txt --no-cache-dir
 ```
 
 ## Usage

data.py

@@ -9,8 +9,6 @@
 from scipy import io
 import random
 
-# seed_everything(42, workers=True)
-
 def _get_file_list(data_path):
     """This function detects all image files within the specified parent
        directory for either training or testing. The path content cannot

download.py

@@ -8,6 +8,7 @@
 from matplotlib.pyplot import imread, imsave
 from scipy.io import loadmat
 from scipy.ndimage import gaussian_filter
+import gdown
 
 def download_salicon(data_path):
     """Downloads the SALICON dataset. Three folders are then created that
@@ -41,15 +42,7 @@ def download_salicon(data_path):
     session = requests.Session()
 
     for count, url in enumerate(urls):
-        response = session.get(url, params={"id": id}, stream=True)
-        token = _get_confirm_token(response)
-
-        if token:
-            params = {"id": id, "confirm": token}
-            response = session.get(url, params=params, stream=True)
-
-        _save_response_content(response, data_path + "tmp.zip")
-
+        gdown.download(url, data_path + "tmp.zip", quiet=False)
         with zipfile.ZipFile(data_path + "tmp.zip", "r") as zip_ref:
             for file in zip_ref.namelist():
                 if "test" not in file:
@@ -417,21 +410,4 @@ def download_fiwi(data_path):
 
     os.remove(data_path + "tmp.zip")
 
-    print("done!", flush=True)
-
-
-def _get_confirm_token(response):
-    for key, value in response.cookies.items():
-        if key.startswith("download_warning"):
-            return value
-
-    return None
-
-def _save_response_content(response, file_path):
-    chunk_size = 32768
-
-    with open(file_path, "wb") as data:
-        for chunk in response.iter_content(chunk_size):
-            if chunk:
-                data.write(chunk)
-
+    print("done!", flush=True)

main.py

@@ -14,10 +14,10 @@
 parser = ArgumentParser()
 parser.add_argument("--learning_rate", type=float, default=0.0001)
 parser.add_argument("--batch_size", type=int, default=64)
-parser.add_argument("--dataset_dir", type=str, default="/home/chakkritt/proj/datasets/")
+parser.add_argument("--dataset_dir", type=str, default="data/")
 parser.add_argument('--input_size_h',default=384, type=int)
 parser.add_argument('--input_size_w',default=384, type=int)
-parser.add_argument('--no_workers',default=16, type=int)
+parser.add_argument('--no_workers',default=8, type=int)
 parser.add_argument('--no_epochs',default=10, type=int)
 parser.add_argument('--log_interval',default=20, type=int)
 parser.add_argument('--lr_sched',default=True, type=bool)

main_ensemble.py

@@ -14,10 +14,10 @@
 parser = ArgumentParser()
 parser.add_argument("--learning_rate", type=float, default=0.0001)
 parser.add_argument("--batch_size", type=int, default=64)
-parser.add_argument("--dataset_dir", type=str, default="/home/mllab/proj/2021/supernet/data/")
+parser.add_argument("--dataset_dir", type=str, default="data/")
 parser.add_argument('--input_size_h',default=256, type=int)
 parser.add_argument('--input_size_w',default=256, type=int)
-parser.add_argument('--no_workers',default=16, type=int)
+parser.add_argument('--no_workers',default=8, type=int)
 parser.add_argument('--no_epochs',default=10, type=int)
 parser.add_argument('--log_interval',default=20, type=int)
 parser.add_argument('--lr_sched',default=True, type=bool)

requirements.txt

@@ -1,8 +1,10 @@
-darmo
-timm
-kornia
-opencv-python
-ptflops
-h5py
-scikit-image
-requests
+darmo==0.1.12
+timm==0.4.12
+kornia==0.6.4
+opencv-python==4.5.5.64
+ptflops==0.6.9
+h5py==3.1.0
+scikit-image==0.17.2
+requests==2.27.1
+matplotlib==3.3.4
+gdown==4.4.0