Dedupe Fails with Tensorflow Error

[NarryG]

I managed to pull my mediaitems down before API deprecation, but when I run the dedupe, it errors out

2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,092: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1086/70030 [00:10<10:33, 108.77it/s]
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,193: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1097/70030 [00:10<10:32, 108.92it/s]
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,300: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1108/70030 [00:10<10:44, 107.02it/s]
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,410: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1119/70030 [00:10<10:57, 104.86it/s]
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,516: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1130/70030 [00:10<10:59, 104.40it/s]
2025-04-01 17:42:40 worker-1  | E0000 00:00:1743543760.529261     314 calculator_graph.cc:898] INVALID_ARGUMENT: CalculatorGraph::Run() failed: 
2025-04-01 17:42:40 worker-1  | Calculator::Process() for node "mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__inferencecalculator__mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__InferenceCalculator" failed: ; RET_CHECK failure (mediapipe/calculators/tensor/inference_calculator_utils.cc:144) (tflite_tensor.bytes)==(input_tensor.bytes())TfLiteTensor and Tensor sizes do not match. TfLiteTensor dims: [1, 224, 224, 3], type: FLOAT32, bytes: 602112 vs. MP Tensor dims: [1, 224, 224, 1], type: kFloat32, bytes: 200704;  at index 0
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,533: WARNING/ForkPoolWorker-31] 
  2%|▏         | 1131/70030 [00:10<10:56, 104.99it/s]
2025-04-01 17:42:40 worker-1  | [2025-04-01 21:42:40,546: ERROR/ForkPoolWorker-31] Task app.tasks.process_duplicates[749a2db5-3c95-4a3a-b92f-3f24d552ddc1] raised unexpected: ValueError('CalculatorGraph::Run() failed: \nCalculator::Process() for node "mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__inferencecalculator__mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__InferenceCalculator" failed: ; RET_CHECK failure (mediapipe/calculators/tensor/inference_calculator_utils.cc:144) (tflite_tensor.bytes)==(input_tensor.bytes())TfLiteTensor and Tensor sizes do not match. TfLiteTensor dims: [1, 224, 224, 3], type: FLOAT32, bytes: 602112 vs. MP Tensor dims: [1, 224, 224, 1], type: kFloat32, bytes: 200704;  at index 0')
2025-04-01 17:42:40 worker-1  | Traceback (most recent call last):
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/lib/duplicate_image_detector.py", line 125, in _calculate_embeddings
2025-04-01 17:42:40 worker-1  |     embedding_result = embedder.embed(mp_image)
2025-04-01 17:42:40 worker-1  |   File "/usr/local/lib/python3.9/site-packages/mediapipe/tasks/python/vision/image_embedder.py", line 229, in embed
2025-04-01 17:42:40 worker-1  |     output_packets = self._process_image_data({
2025-04-01 17:42:40 worker-1  |   File "/usr/local/lib/python3.9/site-packages/mediapipe/tasks/python/vision/core/base_vision_task_api.py", line 95, in _process_image_data
2025-04-01 17:42:40 worker-1  |     return self._runner.process(inputs)
2025-04-01 17:42:40 worker-1  | ValueError: CalculatorGraph::Run() failed: 
2025-04-01 17:42:40 worker-1  | Calculator::Process() for node "mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__inferencecalculator__mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__InferenceCalculator" failed: ; RET_CHECK failure (mediapipe/calculators/tensor/inference_calculator_utils.cc:144) (tflite_tensor.bytes)==(input_tensor.bytes())TfLiteTensor and Tensor sizes do not match. TfLiteTensor dims: [1, 224, 224, 3], type: FLOAT32, bytes: 602112 vs. MP Tensor dims: [1, 224, 224, 1], type: kFloat32, bytes: 200704;  at index 0
2025-04-01 17:42:40 worker-1  | 
2025-04-01 17:42:40 worker-1  | During handling of the above exception, another exception occurred:
2025-04-01 17:42:40 worker-1  | 
2025-04-01 17:42:40 worker-1  | Traceback (most recent call last):
2025-04-01 17:42:40 worker-1  |   File "/usr/local/lib/python3.9/site-packages/celery/app/trace.py", line 453, in trace_task
2025-04-01 17:42:40 worker-1  |     R = retval = fun(*args, **kwargs)
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/__init__.py", line 25, in __call__
2025-04-01 17:42:40 worker-1  |     return self.run(*args, **kwargs)
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/tasks.py", line 111, in process_duplicates
2025-04-01 17:42:40 worker-1  |     results = task_instance.run()
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/lib/process_duplicates_task.py", line 113, in run
2025-04-01 17:42:40 worker-1  |     similarity_map = duplicate_detector.calculate_similarity_map()
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/lib/duplicate_image_detector.py", line 61, in calculate_similarity_map
2025-04-01 17:42:40 worker-1  |     embeddings = self._calculate_embeddings()
2025-04-01 17:42:40 worker-1  |   File "/usr/src/app/app/lib/duplicate_image_detector.py", line 126, in _calculate_embeddings
2025-04-01 17:42:40 worker-1  |     embeddings.append(embedding_result.embeddings[0].embedding)
2025-04-01 17:42:40 worker-1  |   File "/usr/local/lib/python3.9/site-packages/mediapipe/tasks/python/vision/core/base_vision_task_api.py", line 226, in __exit__
2025-04-01 17:42:40 worker-1  |     self.close()
2025-04-01 17:42:40 worker-1  |   File "/usr/local/lib/python3.9/site-packages/mediapipe/tasks/python/vision/core/base_vision_task_api.py", line 209, in close
2025-04-01 17:42:40 worker-1  |     self._runner.close()
2025-04-01 17:42:40 worker-1  | ValueError: CalculatorGraph::Run() failed: 
2025-04-01 17:42:40 worker-1  | Calculator::Process() for node "mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__inferencecalculator__mediapipe_tasks_vision_image_embedder_imageembeddergraph__mediapipe_tasks_core_inferencesubgraph__InferenceCalculator" failed: ; RET_CHECK failure (mediapipe/calculators/tensor/inference_calculator_utils.cc:144) (tflite_tensor.bytes)==(input_tensor.bytes())TfLiteTensor and Tensor sizes do not match. TfLiteTensor dims: [1, 224, 224, 3], type: FLOAT32, bytes: 602112 vs. MP Tensor dims: [1, 224, 224, 1], type: kFloat32, bytes: 200704;  at index 0

Running on Docker Desktop on WSL2, Win11 24H2. I have GPU passthrough set up but it's falling back to CPU if that matters