NICE WORK!!! By the way, upgrade SmallBasicBlock and add STNet module can make better

[zjykzj]

@sirius-ai Thank you very much for implementing LPRNet. During the training process, adding residual connection to SmallBasicBlock using ResNet can further improve performance (LPRNet ->LPRNetPlus); Meanwhile, adding the STNet module can further enhance the evaluation results of CCPD (LPRNet+STNet).

Model	ARCH	Input Shape	GFLOPs	Model Size (MB)	ChineseLicensePlate Accuracy (%)	Training Data	Testing Data
CRNN	CONV+GRU	(3, 48, 168)	4.0	58	82.147	269,621	149,002
CRNN_Tiny	CONV+GRU	(3, 48, 168)	0.3	4.0	76.590	269,621	149,002
LPRNetPlus	CONV	(3, 24, 94)	0.5	2.3	63.546	269,621	149,002
LPRNet	CONV	(3, 24, 94)	0.3	1.9	60.105	269,621	149,002
LPRNetPlus+STNet	CONV	(3, 24, 94)	0.5	2.5	72.130	269,621	149,002
LPRNet+STNet	CONV	(3, 24, 94)	0.3	2.2	72.261	269,621	149,002

The relevant code has been open sourced and can be viewed at zjykzj/crnn-ctc

[wcyjerry]

@zjykzj Amazing guys, I just at the beginning of the task, and I think it have not been update for 6 years, there must be some common method to improve it, and I see your detailed exp result. Great Work!

[wcyjerry]

@zjykzj While, seems Plus+STNet not work well together!