Upload folder using huggingface_hub

convert2PNG.py

@@ -0,0 +1,39 @@
+import argparse
+from PIL import Image
+import json
+import os
+
+
+def process_images(image_filenames):
+    for image_filename in image_filenames:
+        prefix = os.path.splitext(image_filename)[0]
+
+        converted = False
+        if os.path.isfile(image_filename):
+            # 打开图片
+            try:
+                with Image.open(image_filename) as img:
+                    if(img.format != 'PNG'):
+                        img.save(f"{prefix}.png","PNG")
+                        converted=True
+                        print(f"convert {image_filename} to png")
+            except (IOError, SyntaxError) as e:
+                print(e)
+
+            if(converted): os.remove(image_filename)
+        else:
+            print(f"文件 {image_filename} 不存在。")
+
+
+if __name__ == "__main__":
+    # 创建命令行解析器
+    parser = argparse.ArgumentParser(description='读取图片文件并输出为 JSON 格式')
+
+    # 添加位置参数，接受多个图片文件名
+    parser.add_argument('image_filenames', nargs='+', help='要处理的图片文件名')
+
+    # 解析命令行参数
+    args = parser.parse_args()
+
+    # 调用主函数
+    process_images(args.image_filenames)

data/test/bqb.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f4d11e4f5671a4a006e4d79819157917bbf69ea6f4eadc501a11c27276df242
-size 137789440
+oid sha256:1aa0329f2f4e4389f1755f2e1ae94bfe053ad05f1a5cd8110e696818ce086b74
+size 269557760

data/test/nailong.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dda336c6cc2272d7cc8085a5289f3de74b12c0cd34cd008442ebf602c7160649
-size 8386560
+oid sha256:ef9e7c1396472eb1a8e79d4185d02a08c9250f83b57bdcae377e323b33800e31
+size 56064000

data/train/bqb.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:530d6d5bb4216430e466084ba035ee93f3c99c7cc84b43dc17c28460712f8f47
-size 1263370240
+oid sha256:4ef87f29fa9454115b643629cae677d91641fdd9f864b601fd11705878637036
+size 2380390400

data/train/nailong.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c1268667dc31ea55588d79fcad01e1d7cf99098ca21418db9b5f4c4bac44819
-size 80035840
+oid sha256:2ff6b5ee7affa6d45a6519ac304ed6a895c665f0882cbbd4f1d3ce0b8aacfabe
+size 390215680

generate_info.py

@@ -6,6 +6,7 @@ import os
 
 def process_images(image_filenames, label):
     for image_filename in image_filenames:
+        prefix = os.path.splitext(image_filename)[0]
         json_filename = f"{os.path.splitext(image_filename)[0]}.json"
 
         # 检查文件是否存在

tar.sh

@@ -8,10 +8,36 @@
 # @description : tar datas
 ######################################################################
 
-tar   -cvf   ./data/train/bqb.tar      -C ./source/train/bqb  .
-tar   -cvf   ./data/train/nailong.tar   -C ./source/train/nailong .
-tar   -cvf   ./data/test/bqb.tar        -C ./source/test/bqb .
-tar   -cvf   ./data/test/nailong.tar    -C ./source/test/nailong .
+index=(0 1 2 3)
+splits=("train/bqb" "train/nailong" "test/bqb" "test/nailong")
+lables=(0 1 0 1)
+source_dir='source'  # 源目录
+data_dir='data'  # data目录
 
+rm -r $data_dir
+#mkdir $data_dir
+cp -r $source_dir $data_dir
+# 遍历 splits 数组中的每个元素
+for idx in "${index[@]}"; do
+    split=${splits[$idx]}
+    #cp -r "$source_dir/$split" "$data_dir/$splits"
+    folder="$data_dir/$split"
 
+    # 检查目录是否存在
+    if [ -d "$folder" ]; then
+        echo "找到文件夹: $(basename "$folder")"
+        
+        lable=${lables[$idx]}
+        
+        
+        #echo $folder
+        python ./convert2PNG.py "$folder"/*
+
+        python ./generate_info.py "lable=${lables[$idx]}" "$folder"/*
+        
+        tar -cvf "$data_dir/$split.tar" -C "$folder" . && rm -rf "$folder"
+    else
+        echo "目录 $folder 不存在，跳过。"
+    fi
+done