release a prune model, refine code (#158)

slim/nas/eval_nas.py

@@ -26,7 +26,6 @@ import sys
 LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
 SEG_PATH = os.path.join(LOCAL_PATH, "../../", "pdseg")
 sys.path.append(SEG_PATH)
-sys.path.append('/workspace/codes/PaddleSlim1')
 
 import time
 import argparse

slim/prune/README.md

@@ -49,3 +49,10 @@ CUDA_VISIBLE_DEVICES=0
 python -u ./slim/prune/eval_prune.py --cfg configs/cityscape_fast_scnn.yaml --use_gpu --use_mpio \
 TEST.TEST_MODEL your_trained_model \
 ```
+
+## 5. 模型
+
+| 模型 | 数据集合 | 下载地址 |剪裁方法| flops | mIoU on val|
+|---|---|---|---|---|---|
+| Fast-SCNN/bn | Cityscapes |[fast_scnn_cityscapes.tar](https://paddleseg.bj.bcebos.com/models/fast_scnn_cityscape.tar) | 无 | 7.21g | 0.6964 |
+| Fast-SCNN/bn | Cityscapes |[fast_scnn_cityscapes-uniform-51.tar](https://paddleseg.bj.bcebos.com/models/fast_scnn_cityscape-uniform-51.tar) | uniform | 3.54g | 0.6990 |

slim/prune/eval_prune.py

@@ -26,7 +26,6 @@ import sys
 LOCAL_PATH = os.path.dirname(os.path.abspath(__file__))
 SEG_PATH = os.path.join(LOCAL_PATH, "../../", "pdseg")
 sys.path.append(SEG_PATH)
-sys.path.append('/workspace/codes/PaddleSlim1')
 
 import time
 import argparse
@@ -124,7 +123,6 @@ def evaluate(cfg, ckpt_dir=None, use_gpu=False, use_mpio=False, **kwargs):
     if ckpt_dir is not None:
         print('load test model:', ckpt_dir)
         load_model(exe, test_prog, ckpt_dir)
-        #fluid.io.load_params(exe, ckpt_dir, main_program=test_prog)
 
     # Use streaming confusion matrix to calculate mean_iou
     np.set_printoptions(

slim/prune/train_prune.py

@@ -146,11 +146,6 @@ def save_prune_checkpoint(exe, program, ckpt_name):
     if not os.path.isdir(ckpt_dir):
         os.makedirs(ckpt_dir)
 
-    # save_vars(
-    #     exe,
-    #     ckpt_dir,
-    #     program,
-    #     vars=list(filter(fluid.io.is_persistable, program.list_vars())))
     save_model(exe, program, ckpt_dir)
 
     return ckpt_dir
@@ -270,26 +265,9 @@ def train(cfg):
             print_info("Sync BatchNorm strategy will not be effective if GPU device"
                   " count <= 1")
 
-    ####get mobilenetV1 prune parameters####
-    # pruned_params = []
-    # exclude_names = ['aux_layer_lower/logit/weights', 'aux_layer_lower/logit/biases',
-    #                  'aux_layer_higher/logit/weights', 'aux_layer_higher/logit/biases',
-    #                  'classifier/weights', 'classifier/biases']
-    # for x in train_prog.list_vars():
-    #     if isinstance(x, fluid.framework.Parameter):
-    #         # if x.name not in exclude_names:
-    #         if x.name not in exclude_names and "weights" in x.name and "depthwise" not in x.name and "dwise" not in x.name and x.name not in [
-    #             "classifier/dsconv2/pointwise/weights", "classifier/weights", "fc_weights"]:
-    #             pruned_params.append(x.name)
-    #             print(x.name)
-    # print('to prune paramter number:', len(pruned_params))
-    # print('listtttt:', ','.join(pruned_params))
-    # pruned_ratios = [0.1] * len(pruned_params)
-    ########################################
     pruned_params = cfg.SLIM.PRUNE_PARAMS.strip().split(',')
-    #pruned_params = [str(x) for x in pruned_params]
-    print('paramssss:', pruned_params)
     pruned_ratios = cfg.SLIM.PRUNE_RATIOS
+
     if isinstance(pruned_ratios, float):
         pruned_ratios = [pruned_ratios] * len(pruned_params)
     elif isinstance(pruned_ratios, (list, tuple)):