模型难以作为实时视频，有没有什么办法做图片编码预处理或者设置？

[waka19910909]

A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.USER: <media><media><media><media><media>查看这组图片序列中行人的活动变化，分析是否发生事件(如：人员摔倒，人员聚集在电梯口等)；结果使用一句话回答，不超过30个字ASSISTANT:
encoding image slice...
image slice encoded in 443 ms
decoding image batch 1/1, n_tokens_batch = 256
image decoded (batch 1/1) in 113 ms
encoding image slice...
image slice encoded in 402 ms
decoding image batch 1/1, n_tokens_batch = 256
image decoded (batch 1/1) in 123 ms
encoding image slice...
image slice encoded in 402 ms
decoding image batch 1/1, n_tokens_batch = 256
image decoded (batch 1/1) in 155 ms
encoding image slice...
image slice encoded in 358 ms
decoding image batch 1/1, n_tokens_batch = 256
image decoded (batch 1/1) in 165 ms
encoding image slice...
image slice encoded in 361 ms
decoding image batch 1/1, n_tokens_batch = 256
image decoded (batch 1/1) in 179 ms
Llama.generate: 1363 prefix-match hit, remaining 1 prompt tokens to eval
llama_perf_context_print: load time = 26.42 ms
llama_perf_context_print: prompt eval time = 3458.26 ms / 1365 tokens ( 2.53 ms per token, 394.71 tokens per second)
llama_perf_context_print: eval time = 931.93 ms / 36 runs ( 25.89 ms per token, 38.63 tokens per second)
llama_perf_context_print: total time = 1046.71 ms / 1401 tokens
分析结果: 从图片中可以看到，有几个人正在移动。其中一个人正在从电梯口出来，其他人则在周围行走。没有看到人员摔倒或聚集在电梯口等事件。
批处理时间: 5.04秒, 帧率: 0.99帧/秒
使用边缘计算盒子部署模型推理时，发现每一张图片在推理时都需要编码并耗费大量的时间（可以说是主要时间）