"Welcome to the object detection colab! This demo will take you through the steps of running an \"out-of-the-box\" detection model on a collection of images."
]
},
{
"cell_type": "markdown",
"metadata": {
"colab_type": "text",
"id": "vPs64QA1Zdov"
},
"source": [
"## Imports and Setup"
]
},
{
"cell_type": "code",
"metadata": {
"id": "LBZ9VWZZFUCT",
"colab_type": "code",
"colab": {}
},
"source": [
"!pip install -U --pre tensorflow==\"2.2.0\""
],
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "oi28cqGGFWnY",
"colab_type": "code",
"colab": {}
},
"source": [
"import os\n",
"import pathlib\n",
"\n",
"# Clone the tensorflow models repository if it doesn't already exist\n",
"if \"models\" in pathlib.Path.cwd().parts:\n",
" while \"models\" in pathlib.Path.cwd().parts:\n",
"Label maps correspond index numbers to category names, so that when our convolution network predicts `5`, we know that this corresponds to `airplane`. Here we use internal utility functions, but anything that returns a dictionary mapping integers to appropriate string labels would be fine."
"Run the below code which loads an image, runs it through the detection model and visualizes the detection results.\n",
"\n",
"Note that this will take a long time (several minutes) the first time you run this code due to tf.function's trace-compilation --- on subsequent runs (e.g. on new images), things will be faster.\n",
"\n",
"Here are some simple things to try out if you are curious:\n",
"* Try running inference on your own images (local paths work)\n",
"* Modify some of the input images and see if detection still works. Some simple things to try out here (just uncomment the relevant portions of code) include flipping the image horizontally, or converting to grayscale (note that we still expect the input image to have 3 channels).\n",
"* Print out `detections['detection_boxes']` and try to match the box locations to the boxes in the image. Notice that coordinates are given in normalized form (i.e., in the interval [0, 1]).\n",
"* Set min_score_thresh to other values (between 0 and 1) to allow more detections in or to filter out more detections.\n",
"\n",
"Note that you can run this cell repeatedly without rerunning earlier cells.\n"
"## Digging into the model's intermediate predictions\n",
"\n",
"For this part we will assume that the detection model is a CenterNet model following Zhou et al (https://arxiv.org/abs/1904.07850). And more specifically, we will assume that `detection_model` is of type `meta_architectures.center_net_meta_arch.CenterNetMetaArch`.\n",
"\n",
"As one of its intermediate predictions, CenterNet produces a heatmap of box centers for each class (for example, it will produce a heatmap whose size is proportional to that of the image that lights up at the center of each, e.g., \"zebra\"). In the following, we will visualize these intermediate class center heatmap predictions."
