Object detection with TFLite

I need to convert a given tensorflow model into a tflite model. Unfortunately I am not allowed to share the model. I also have not created it myself. However, I know it is a "ssdmobilenet_v2_320x320" model and the object detection api was used to train the model. Right now the model is used to detect a single class.

When I run the original model the results are very good. However, when I use the tflite model, the results feel kinda random. All the scores of the ten detected objects are nearly zero.

I used this google colab notebook for converting the model. I experimented with converter.experimental_new_converter = True and converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS,.lite.OpsSet.SELECT_TF_OPS] but it didn't help.

Here is the meta data output of the tflite model.

Metadata populated:
{
  "name": "ObjectDetector",
  "description": "Identify which of a known set of objects might be present and provide information about their positions within the given image or a video stream.",
  "subgraph_metadata": [
    {
      "input_tensor_metadata": [
        {
          "name": "image",
          "description": "Input image to be detected.",
          "content": {
            "content_properties_type": "ImageProperties",
            "content_properties": {
              "color_space": "RGB"
            }
          },
          "process_units": [
            {
              "options_type": "NormalizationOptions",
              "options": {
                "mean": [
                  127.5
                ],
                "std": [
                  127.5
                ]
              }
            }
          ],
          "stats": {
            "max": [
              1.0
            ],
            "min": [
              -1.0
            ]
          }
        }
      ],
      "output_tensor_metadata": [
        {
          "name": "location",
          "description": "The locations of the detected boxes.",
          "content": {
            "content_properties_type": "BoundingBoxProperties",
            "content_properties": {
              "index": [
                1,
                0,
                3,
                2
              ],
              "type": "BOUNDARIES"
            },
            "range": {
              "min": 2,
              "max": 2
            }
          },
          "stats": {
          }
        },
        {
          "name": "category",
          "description": "The categories of the detected boxes.",
          "content": {
            "content_properties_type": "FeatureProperties",
            "content_properties": {
            },
            "range": {
              "min": 2,
              "max": 2
            }
          },
          "stats": {
          },
          "associated_files": [
            {
              "name": "tflite_label_map.txt",
              "description": "Labels for categories that the model can recognize.",
              "type": "TENSOR_VALUE_LABELS"
            }
          ]
        },
        {
          "name": "score",
          "description": "The scores of the detected boxes.",
          "content": {
            "content_properties_type": "FeatureProperties",
            "content_properties": {
            },
            "range": {
              "min": 2,
              "max": 2
            }
          },
          "stats": {
          }
        },
        {
          "name": "number of detections",
          "description": "The number of the detected boxes.",
          "content": {
            "content_properties_type": "FeatureProperties",
            "content_properties": {
            }
          },
          "stats": {
          }
        }
      ],
      "output_tensor_groups": [
        {
          "name": "detection_result",
          "tensor_names": [
            "location",
            "category",
            "score"
          ]
        }
      ]
    }
  ],
  "min_parser_version": "1.2.0"
}

=============================
Associated file(s) populated:
['tflite_label_map.txt']

I actually don't really know how to interpret most of it since I didn't find any documentation for the metadata. But the lines

 "range": {
          "min": 2,
          "max": 2
        }

feel wrong. Which ranges are those and why should the min and the max be 2?