Hi Dat, Thanks for the article. Very helpful.
I was trying to follow the post in medium to create my own object detection library, for classifying car models.
I tried implementing it for a car type classification — Trying to classify different kinds of coupe of a brand. I started with 2 types. When I try to classify my image, it always shows as type 1, though image is of type 2. Can someone pls help me? It is mostly what I got from the your github, with tweaks for multiple classes.
Here is my github repo url — https://github.com/iosnewbie2016/TFObjectDetection.
I had put only some images and couple of bounding label xmls for each class, for simplicity. Here is brief info abt my files.
custom_object_detection.ipynb — will consolidate all Xmls to a single csv for training and testing Updated the label map in gegnerate_tfrecord.py script. Run the generate_tfrecord.py script python3 generate_tfrecord.py — csv_input=data/train_labels.csv — output_path=data/train.record python3 generate_tfrecord.py — csv_input=data/test_labels.csv — output_path=data/test.record
Am using ssd_mobilenet_v1_coco_11_06_2017 & ssd_mobilenet_v1_pets.config. It is present in training directory. Ran only 750 iterations.
python3 train.py — logtostderr — train_dir=training/ — pipeline_config_path=training/ssd_mobilenet_v1_pets.config
Run the below to create the graph.
python3 export_inference_graph.py — input_type image_tensor — pipeline_config_path training/ssd_mobilenet_v1_pets.config — trained_checkpoint_prefix training/model.ckpt-750 — output_directory C63AMG_inference_graph
Can you pls help? Thanks for your time.