fix: operator class names and imports

.github/workflows/benchmarks.yml

@@ -7,46 +7,46 @@ jobs:
   benchmark:
     runs-on: ubuntu-latest
     timeout-minutes: 120
-    
+
     steps:
       - uses: actions/checkout@v4
-        
+
       - name: Install uv
         uses: astral-sh/setup-uv@v5
-        
+
       - name: Install system dependencies
         run: |
           sudo apt-get update
           sudo apt-get install -y ffmpeg tesseract-ocr tesseract-ocr-hin tesseract-ocr-tam tesseract-ocr-tel
-          
+
       - name: Setup Python environment
         run: |
           uv venv
           source .venv/bin/activate
           uv pip install ".[all]" ".[dev]"
           uv pip install gdown
-      
+
       - name: Download benchmark videos
         run: |
           gdown --folder "https://drive.google.com/drive/folders/1XozZbqSzJVzTD4QgGZNOR6U5U7_rf5O3" -O benchmark/videos
-          
+
       - name: Run benchmarks
         run: |
           source .venv/bin/activate
           cd benchmark
           python main.py
-          
+
       - name: Upload benchmark results
         uses: actions/upload-artifact@v4
         with:
           name: benchmark-results-${{ github.run_id }}
           path: benchmark/results/
           retention-days: 90
-          
+
       - name: Cleanup caches
         if: always()
         run: |
           rm -rf ~/.cache/torch/hub
           rm -rf ~/.cache/huggingface/hub
           rm -rf benchmark/videos
-          rm -rf benchmark/images
+          rm -rf benchmark/images

.gitignore

@@ -158,4 +158,4 @@ src/core/operators/cnn/panns_data
 # uv
 uv.lock
 benchmark/images/
-benchmark/videos/
+benchmark/videos/

README.md

@@ -56,10 +56,10 @@ operators :
   label : "Operators"
   parameters :
     - name : "Video Vector Representation"
-      type : "vid_vec_rep_clip"
+      type : "vid_vec_rep"
       parameters: {}
     - name : "Image Vector Representation"
-      type : "image_vec_rep_resnet"
+      type : "image_vec_rep"
       parameters: {}
 ```
 
@@ -82,7 +82,7 @@ feluda = Feluda(config_path)
 feluda.setup()
 
 # Access an operator and run a task
-operator = feluda.operators.get()["vid_vec_rep_clip"]
+operator = feluda.operators.get()["vid_vec_rep"]
 result = operator.run("path/to/example.mp4")
 print(result)
 ```

benchmark/README.md

@@ -96,7 +96,7 @@ def benchmark() -> list[dict]:
     test_data = [
         {"file": "test/image1.jpg", "labels": ["label1", "label2"]},
         {"file": "test/image2.jpg", "labels": ["label1", "label3"]},
-    ]      
+    ]
     for test_item in test_data:
         # Profile the operator for each test_item and add it to results.
         result = Profiler.benchmark_operator(
@@ -105,8 +105,8 @@ def benchmark() -> list[dict]:
             runtime_kwargs=test_item,
             operator_kwargs=
         )
-        
-        results.append(result)   
+
+        results.append(result)
     return results
 ```
 
@@ -166,8 +166,8 @@ Provides human-readable summaries with:
 pip install psutil memory_profiler numpy pillow
 ```
 
-or 
+or
 
 ```bash
 pip install feluda['dev']
-```
+```

benchmark/operators/__init__.py

@@ -2,20 +2,20 @@
     # classify_video_zero_shot,
     cluster_embeddings,
     detect_lewd_images,
-    detect_text_in_image_tesseract,
+    detect_text_in_image,
     dimension_reduction,
-    image_vec_rep_resnet,
-    vid_vec_rep_clip,
-    video_hash_tmk,
+    image_vec_rep,
+    vid_vec_rep,
+    video_hash,
 )
 
 all_operators = [
     # classify_video_zero_shot,
     cluster_embeddings,
     detect_lewd_images,
-    detect_text_in_image_tesseract,
+    detect_text_in_image,
     dimension_reduction,
-    image_vec_rep_resnet,
-    vid_vec_rep_clip,
-    video_hash_tmk,
+    image_vec_rep,
+    vid_vec_rep,
+    video_hash,
 ]

benchmark/operators/classify_video_zero_shot.py

@@ -1,11 +1,11 @@
 from benchmark.config import BENCHMARK_VIDEOS
 from benchmark.profiler import Profiler
 from feluda.factory import VideoFactory
-from operators.classify_video_zero_shot import VideoClassifier
+from operators.classify_video_zero_shot import ClassifyVideoZeroShot
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the VideoClassifier operator."""
+    """Benchmark the ClassifyVideoZeroShot operator."""
     test_data = [
         {"file": VideoFactory.make_from_file_on_disk(video_path)}
         for video_path in BENCHMARK_VIDEOS
@@ -14,7 +14,7 @@ def benchmark() -> list[dict]:
     for test_item in test_data:
         print(test_item)
         result = Profiler.benchmark_operator(
-            operator_class=VideoClassifier,
+            operator_class=ClassifyVideoZeroShot,
             operator_name="video_classifier",
             runtime_kwargs=test_item,
         )

benchmark/operators/detect_lewd_images.py

@@ -1,16 +1,16 @@
 from benchmark.config import BENCHMARK_IMAGES
 from benchmark.profiler import Profiler
-from operators.detect_lewd_images import LewdImageDetector
+from operators.detect_lewd_images import DetectLewdImages
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the LewdImageDetector operator."""
+    """Benchmark the DetectLewdImages operator."""
     test_data = [{"file": image} for image in BENCHMARK_IMAGES]
 
     results = []
     for test_item in test_data:
         result = Profiler.benchmark_operator(
-            operator_class=LewdImageDetector,
+            operator_class=DetectLewdImages,
             operator_name="detect_lewd_images",
             runtime_kwargs=test_item,
         )

benchmark/operators/detect_text_in_image_tesseract.py

@@ -1,17 +1,17 @@
 from benchmark.config import BENCHMARK_IMAGES
 from benchmark.profiler import Profiler
-from operators.detect_text_in_image_tesseract import ImageTextDetector
+from operators.detect_text_in_image import DetectTextInImage
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the ImageTextDetector operator."""
+    """Benchmark the DetectTextInImage operator."""
     test_data = [{"file": image} for image in BENCHMARK_IMAGES]
 
     results = []
     for test_item in test_data:
         result = Profiler.benchmark_operator(
-            operator_class=ImageTextDetector,
-            operator_name="detect_text_in_image_tesseract",
+            operator_class=DetectTextInImage,
+            operator_name="detect_text_in_image",
             runtime_kwargs=test_item,
         )
         results.append(result)

benchmark/operators/image_vec_rep_resnet.py

@@ -1,11 +1,11 @@
 from benchmark.config import BENCHMARK_IMAGES
 from benchmark.profiler import Profiler
 from feluda.factory import ImageFactory
-from operators.image_vec_rep_resnet import ImageVecRepResnet
+from operators.image_vec_rep import ImageVecRep
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the ImageVecRepResnet operator."""
+    """Benchmark the ImageVecRep operator."""
     test_data = [
         {"image": ImageFactory.make_from_file_on_disk_to_path(image)}
         for image in BENCHMARK_IMAGES
@@ -14,8 +14,8 @@ def benchmark() -> list[dict]:
     results = []
     for test_item in test_data:
         result = Profiler.benchmark_operator(
-            operator_class=ImageVecRepResnet,
-            operator_name="image_vec_rep_resnet",
+            operator_class=ImageVecRep,
+            operator_name="image_vec_rep",
             runtime_kwargs=test_item,
         )
         results.append(result)

benchmark/operators/vid_vec_rep_clip.py

@@ -1,11 +1,11 @@
 from benchmark.config import BENCHMARK_VIDEOS
 from benchmark.profiler import Profiler
 from feluda.factory import VideoFactory
-from operators.vid_vec_rep_clip import VidVecRepClip
+from operators.vid_vec_rep import VidVecRep
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the VidVecRepClip operator."""
+    """Benchmark the VidVecRep operator."""
     test_data = [
         {"file": VideoFactory.make_from_file_on_disk(video_path)}
         for video_path in BENCHMARK_VIDEOS
@@ -14,8 +14,8 @@ def benchmark() -> list[dict]:
 
     for test_item in test_data:
         result = Profiler.benchmark_operator(
-            operator_class=VidVecRepClip,
-            operator_name="vid_vec_rep_clip",
+            operator_class=VidVecRep,
+            operator_name="vid_vec_rep",
             runtime_kwargs=test_item,
         )
         results.append(result)

benchmark/operators/video_hash_tmk.py

@@ -1,17 +1,17 @@
 from benchmark.config import BENCHMARK_VIDEOS
 from benchmark.profiler import Profiler
-from operators.video_hash_tmk import VideoHashTmk
+from operators.video_hash import VideoHash
 
 
 def benchmark() -> list[dict]:
-    """Benchmark the VideoHashTmk operator."""
+    """Benchmark the VideoHash operator."""
     test_data = [{"video_path": video_path} for video_path in BENCHMARK_VIDEOS]
 
     results = []
     for test_item in test_data:
         result = Profiler.benchmark_operator(
-            operator_class=VideoHashTmk,
-            operator_name="video_hash_tmk",
+            operator_class=VideoHash,
+            operator_name="video_hash",
             runtime_kwargs=test_item,
         )
         results.append(result)

benchmark/results/benchmark_results_20250829.json

@@ -1033,4 +1033,4 @@
       "success_rate": 1.0
     }
   }
-}
+}

benchmark/results/benchmark_results_20250829.md

@@ -18,7 +18,7 @@ Generated: 2025-08-28T23:55:24.834470
 | cluster_embeddings_dbscan | 3 | 100.0 | 6.256 | 6.214 | 6.277 | 0.030 | -10.64 | 851.92 |
 | cluster_embeddings_kmeans | 3 | 100.0 | 12.851 | 10.772 | 14.850 | 1.666 | -38.10 | 846.00 |
 | detect_lewd_images | 11 | 100.0 | 8.863 | 6.579 | 9.632 | 0.817 | -18.47 | 1905.08 |
-| detect_text_in_image_tesseract | 11 | 100.0 | 9.381 | 5.869 | 15.024 | 2.735 | 0.54 | 1615.25 |
+| detect_text_in_image | 11 | 100.0 | 9.381 | 5.869 | 15.024 | 2.735 | 0.54 | 1615.25 |
 | dimension_reduction_tsne | 4 | 100.0 | 30.234 | 6.242 | 63.125 | 21.633 | 40.23 | 2008.08 |
 | dimension_reduction_umap | 4 | 100.0 | 12.339 | 8.190 | 16.055 | 3.177 | 18.89 | 2041.16 |
 
@@ -46,17 +46,17 @@ Generated: 2025-08-28T23:55:24.834470
 | detect_lewd_images | - | success | 9.087 | -21.03 | 1848.25 | 1.734 |  |
 | detect_lewd_images | - | success | 6.579 | 53.12 | 1865.45 | 1.780 |  |
 | detect_lewd_images | - | success | 9.302 | -24.25 | 1877.45 | 1.800 |  |
-| detect_text_in_image_tesseract | - | success | 12.963 | 5.95 | 1615.22 | 0.068 |  |
-| detect_text_in_image_tesseract | - | success | 10.756 | 0.03 | 1615.25 | 0.063 |  |
-| detect_text_in_image_tesseract | - | success | 15.024 | 0.00 | 1615.25 | 0.062 |  |
-| detect_text_in_image_tesseract | - | success | 11.240 | 0.00 | 1615.25 | 0.080 |  |
-| detect_text_in_image_tesseract | - | success | 6.466 | 0.00 | 1615.25 | 0.060 |  |
-| detect_text_in_image_tesseract | - | success | 8.662 | 0.00 | 1615.25 | 0.054 |  |
-| detect_text_in_image_tesseract | - | success | 8.693 | 0.00 | 1615.25 | 0.053 |  |
-| detect_text_in_image_tesseract | - | success | 6.450 | 0.00 | 1615.25 | 0.101 |  |
-| detect_text_in_image_tesseract | - | success | 5.869 | 0.00 | 1615.25 | 0.059 |  |
-| detect_text_in_image_tesseract | - | success | 8.997 | 0.00 | 1615.25 | 0.093 |  |
-| detect_text_in_image_tesseract | - | success | 8.067 | 0.00 | 1615.25 | 0.055 |  |
+| detect_text_in_image | - | success | 12.963 | 5.95 | 1615.22 | 0.068 |  |
+| detect_text_in_image | - | success | 10.756 | 0.03 | 1615.25 | 0.063 |  |
+| detect_text_in_image | - | success | 15.024 | 0.00 | 1615.25 | 0.062 |  |
+| detect_text_in_image | - | success | 11.240 | 0.00 | 1615.25 | 0.080 |  |
+| detect_text_in_image | - | success | 6.466 | 0.00 | 1615.25 | 0.060 |  |
+| detect_text_in_image | - | success | 8.662 | 0.00 | 1615.25 | 0.054 |  |
+| detect_text_in_image | - | success | 8.693 | 0.00 | 1615.25 | 0.053 |  |
+| detect_text_in_image | - | success | 6.450 | 0.00 | 1615.25 | 0.101 |  |
+| detect_text_in_image | - | success | 5.869 | 0.00 | 1615.25 | 0.059 |  |
+| detect_text_in_image | - | success | 8.997 | 0.00 | 1615.25 | 0.093 |  |
+| detect_text_in_image | - | success | 8.067 | 0.00 | 1615.25 | 0.055 |  |
 | dimension_reduction_tsne | small_500x512 - tsne | success | 6.242 | 5.47 | 1679.20 | 8.336 |  |
 | dimension_reduction_umap | small_500x512 - umap | success | 8.190 | 10.77 | 1689.98 | 2.457 |  |
 | dimension_reduction_tsne | medium_2000x512 - tsne | success | 16.433 | 22.95 | 1712.94 | 85.710 |  |

docs/api/operators.md

@@ -6,11 +6,11 @@ This section documents all the available operators in Feluda.
 :maxdepth: 1
 :caption: Operators
 
-operators/image_vec_rep_resnet
+operators/image_vec_rep
 operators/detect_lewd_images
-operators/detect_text_in_image_tesseract
-operators/vid_vec_rep_clip
-operators/video_hash_tmk
+operators/detect_text_in_image
+operators/vid_vec_rep
+operators/video_hash
 operators/classify_video_zero_shot
 operators/cluster_embeddings
 operators/dimension_reduction

docs/api/operators/classify_video_zero_shot.md

@@ -3,7 +3,7 @@
 ```
 
 ```{eval-rst}
-.. autoclass:: operators.classify_video_zero_shot.classify_video_zero_shot.VideoClassifier
+.. autoclass:: operators.classify_video_zero_shot.classify_video_zero_shot.ClassifyVideoZeroShot
    :members:
    :undoc-members:
    :show-inheritance:

docs/api/operators/detect_lewd_images.md

@@ -3,7 +3,7 @@
 ```
 
 ```{eval-rst}
-.. autoclass:: operators.detect_lewd_images.detect_lewd_images.LewdImageDetector
+.. autoclass:: operators.detect_lewd_images.detect_lewd_images.DetectLewdImages
    :members:
    :undoc-members:
    :show-inheritance:

docs/api/operators/detect_text_in_image_tesseract.md

@@ -1,9 +1,9 @@
-```{include} ../../../operators/detect_text_in_image_tesseract/README.md
+```{include} ../../../operators/detect_text_in_image/README.md
 :parser: myst_parser.sphinx_
 ```
 
 ```{eval-rst}
-.. autoclass:: operators.detect_text_in_image_tesseract.detect_text_in_image_tesseract.ImageTextDetector
+.. autoclass:: operators.detect_text_in_image.detect_text_in_image.DetectTextInImage
    :members:
    :undoc-members:
    :show-inheritance:

docs/api/operators/image_vec_rep_resnet.md

@@ -1,9 +1,9 @@
-```{include} ../../../operators/image_vec_rep_resnet/README.md
+```{include} ../../../operators/image_vec_rep/README.md
 :parser: myst_parser.sphinx_
 ```
 
 ```{eval-rst}
-.. autoclass:: operators.image_vec_rep_resnet.image_vec_rep_resnet.ImageVecRepResnet
+.. autoclass:: operators.image_vec_rep.image_vec_rep.ImageVecRep
    :members:
    :undoc-members:
    :show-inheritance:

docs/api/operators/vid_vec_rep_clip.md

@@ -1,9 +1,9 @@
-```{include} ../../../operators/vid_vec_rep_clip/README.md
+```{include} ../../../operators/vid_vec_rep/README.md
 :parser: myst_parser.sphinx_
 ```
 
 ```{eval-rst}
-.. autoclass:: operators.vid_vec_rep_clip.vid_vec_rep_clip.VidVecRepClip
+.. autoclass:: operators.vid_vec_rep.vid_vec_rep.VidVecRep
    :members:
    :undoc-members:
    :show-inheritance: