First of all introduce myself
Something about the project (clear)
For project ask questions (in the project is responsible for which block, if algorithm, tell me the algorithm) feel algorithm (models) can also speak, speak (algorithm) shoes seal is bad, not highlight own workload,
Shoes print project: responsible for which block algorithm, identification, use what algorithm, (back to prepare for next)
Models project: CNN, fitting problem (the number of data quantity is little, with few) note see ImagNet data set, each type of about how much the amount of data, the data quantity is little, how much will happen fitting)
Accuracy and recall (recognition) top1 top5
Next time you can say a ha, only on the commonly used network top1 top5 recall, and say your accuracy classification accuracy of the algorithm and how to calculate the confusion matrix, the AUC value, recall, respectively, how to calculate?
How to expand the data set?
How the data increase?
Model points wrong what, why do wrong? What's the solution?
Classification models under complex environment, in the evening, fog, etc., how to deal with? (should belong to the data enhancement)
Familiarity with programming language?
Results?
CodePudding user response:
Thanks for sharing the
CodePudding user response:
Support the building Lord