The use of tensorflow object_detection API to train their own data encountered and solutions

1.Windows not found appears under object_detection problems package .

Solution

　　In Anaconda3 \ soft \ Lib \ site- packages create a pth file, PedestrianDetection folders and slim folder path is written.

2. Training in emerging issues

① raise ValueError('First step cannot be zero.')ValueError: First step cannot be zero.

　　Delete the pre-training model pipeline.Config the following code

schedule{
        step:0,
        learning_rate:0.00008797
}

 

 ②Assertion failed: [maximum box coordinate value is larger than 1.100000: ] [1.15277779]

Question: annotation data beyond the length and width of the image

Solution: When generating annotation data, the x, y coordinates truncate