CODEWITHSUNDEEP
arraystensorflowobject-detectiontensorflow-litetoco
Marwen Bhj
Marwen Bhj

Reputation: 67

tflite object detection inference not working properly

bazel run -c opt tensorflow/lite/toco:toco -- \
--input_file=$INPUT_PATH/tflite_graph.pb \
--output_file=$OUTPUT_PATH/detect.tflite \
--input_shapes=1,300,300,3 \
--input_arrays=normalized_input_image_tensor \
--output_arrays='TFLite_Detection_PostProcess','TFLite_Detection_PostProcess:1','TFLite_Detection_PostProcess:2','TFLite_Detection_PostProcess:3' \
--inference_type=FLOAT \
--allow_custom_ops

Describe the problem

I re-trained a float model ssd_mobilenet_v2_coco on my own dataset to detect one object , after freezing the graph, and runing the inference using this notebook, the model performed well and as expected. Afterwards, I exported the tflite_graph.pb using this command : 

python3 object_detection/export_tflite_ssd_graph.py \
--pipeline_config_path=/PATH/pipeline.config \
--trained_checkpoint_prefix=/PATH/model.ckpt-50000 \
--output_directory=/PATH/tflite \
--add_postprocessing_op=True

Then converted the tflite_graph.pb into a detect.tflite using this command : 

bazel run -c opt tensorflow/lite/toco:toco -- \
--input_file=/PATH/tflite_graph.pb \
--output_file=/PATH/detect.tflite \
--input_shapes=1,300,300,3 \
--input_arrays=normalized_input_image_tensor \
--output_arrays='TFLite_Detection_PostProcess','TFLite_Detection_PostProcess:1','TFLite_Detection_PostProcess:2','TFLite_Detection_PostProcess:3' \
--inference_type=FLOAT \
--allow_custom_ops

The labelmap.txt contains the following label using the coco dataset labelmap style : 

???
Object_1

I run the inference using this script which is a custom version of this script

The output is pretty odd as we have negative values and inaccurate ones and therefore we can't visualize any detection : 

image: /PATH/Inference_Notebooks/test_images/image15.jpg
boxes: 
[[  3.1171997    4.9266524  -15.893956     7.4959326 ]
 [ -1.904324     1.0337602   -7.818109    -7.9575577 ]
 [  1.4761205    2.4604938  -14.553953     8.159015  ]
 [  3.4024968    2.7483184   -9.744125     6.332321  ]
 [ -4.447262    -2.6578145   -1.9118335  -12.579478  ]
 [  1.5781038   -2.980986   -15.902752     5.9894757 ]
 [ -0.4003443  -12.641836    -5.6216025   -0.9522631 ]
 [ -1.3472033   -5.514964    -4.7609305  -11.9099045 ]
 [  2.6661258   -4.2592344  -13.687805    -4.15193   ]
 [ -0.49181542   9.271766    -3.5316777   -3.233222  ]] 
classes: [  2   0 -10   9  -4   3  -6  -7   0   2] 
scores: [ -5.54325      1.9508497   -6.1604195   -4.2281013   -0.02703065 0.707987   -11.534006     7.781439    -2.5938861   -2.5299745 ] 
number of detections: 0.0

I tested the inference files on default SSD mobilenet models using their default weights trained on the COCO dataset, and I was able to visualize the boxes, as the model detected cars, people, etc.. I converted a QUANTIZED tflite_graph.pb of defualt models like the quantized SSD mobilenet v1 into detect.tflite, and it was able to output boxes and the coco dataset labels.

I don't understand where these values are coming from, and why the tflite model isn't detecting while the tensorflow one does .

Also a quantized SSD mobilenetv2 model I trained and converted following the same steps but using this command: 

bazel run -c opt tensorflow/lite/toco:toco -- \
--input_file=/PATH/tflite_graph.pb \
--output_file=/PATH/detect.tflite \
--input_shapes=1,300,300,3 \
--input_arrays=normalized_input_image_tensor \
--output_arrays='TFLite_Detection_PostProcess','TFLite_Detection_PostProcess:1','TFLite_Detection_PostProcess:2','TFLite_Detection_PostProcess:3' \
--inference_type=QUANTIZED_UINT8 \
--mean_values=128 \
--std_values=128 \
--change_concat_input_ranges=false \
--allow_custom_ops

outputs these values: 

image: /home/neox/HistorIAR_Detection/Inference_Notebooks/test_images/image15.jpg 
boxes: 
[[1.0170084e+32 1.6485735e+30 2.5740889e+31 2.6057175e+31]
 [2.4791379e+31 6.3874716e+33 1.0232719e+32 1.0043315e+32]
 [6.4686013e+33 4.0425799e+32 1.0107439e+32 2.5712148e+34]
 [1.6069700e+33 4.0430743e+32 2.5712782e+34 1.0106698e+32]
 [2.5426435e+31 1.0233461e+32 1.0232968e+32 1.0170082e+32]
 [1.6272522e+33 4.0426789e+32 1.0234205e+32 1.6272126e+33]
 [2.5266129e+31 6.5147562e+30 2.5740879e+31 2.5742122e+31]
 [1.0423612e+32 1.0296598e+32 6.5144491e+30 6.3561451e+30]
 [1.0170081e+32 1.6372740e+33 6.1586925e+30 1.6170719e+33]
 [4.0172261e+32 1.0170823e+32 6.5090083e+33 1.0106451e+32]] 
classes: [-2147483648 -2147483648 -2147483648 -2147483648 -2147483648 -2147483648
 -2147483648 -2147483648 -2147483648 -2147483648] 
scores: [6.3787720e+33 1.0090909e+35 6.3066602e+33 1.6144633e+36 1.6146259e+36 1.6042820e+36 1.6145852e+36 6.4585042e+36 1.6042415e+36 4.0624248e+35] 
num: 0.0

Upvotes: 4

Views: 1844

Answers (2)

cml
cml

Reputation: 131

The output of tflite model requires post-processing. The model returns a fixed number (here, 10 detections) by default. Use the output tensor at index 3 to get the number of valid boxes, num_det. (i.e. top num_det detections are valid, ignore the rest).

num_det = int(interpreter.get_tensor(output_details[3]['index']))
boxes = interpreter.get_tensor(output_details[0]['index'])[0][:num_det]
classes = interpreter.get_tensor(output_details[1]['index'])[0][:num_det]
scores = interpreter.get_tensor(output_details[2]['index'])[0][:num_det]

As for your question, there is no valid detections num: 0.0. So the output tensors have garbage values.

Here's a link to an inference script with input preprocessing, output post-processing and mAP evaluation.

Upvotes: 2

Rajneesh Verma
Rajneesh Verma

Reputation: 1

   `frame=cv2.resize(frame,(300,300))
    frame=np.expand_dims(frame,axis=0)
    frame=frame/128.0
    frame=frame.astype('float64')
    frame=np.interp(frame,(frame.min(),frame.max()),(-1,+1))

last line np.interp is very important for ssd models

Upvotes: 0

Related Questions