试试这个。我创建了这个表：

CREATE TABLE `test_data2` (
  `id` bigint unsigned NOT NULL AUTO_INCREMENT,
  `company_id` int unsigned NOT NULL,
  `image_path` varchar(50) COLLATE utf8mb4_unicode_ci NOT NULL,
  PRIMARY KEY (`id`)
);

数据：

id  company_id  image_path
------  ----------  ------------
     1         100  img_100_a
     2         100  img_100_b
     3         100  img_100_c
     4         100  img_100_d
     5         101  img_101_a
     6         101  img_101_b

和以下查询：

SELECT id, company_id, image_path,
    (
        CASE company_id 
        WHEN @cur_company_id
        THEN @curRow := @curRow + 1 
        ELSE @curRow := 1 AND @cur_company_id := company_id
        END
    ) AS company_image_counter
FROM test_data2
ORDER BY company_id

给出输出：

id  company_id  image_path  company_image_counter
------  ----------  ----------  -----------------------
     1         100  img_100_a                         1
     2         100  img_100_b                         2
     3         100  img_100_c                         3
     4         100  img_100_d                         4
     5         101  img_101_a                         1
     6         101  img_101_b                         2

因此，对于公司ID 100，对于4个图像，它给出计数器1到4，对于公司101，对于2个图像，它给出计数器1到2。
上面的代码也可以在旧版本的MySQL中运行。如果你有MySQL 8+，它有LAG和LEAD函数。我想看看这些东西是否能帮上忙。
更新：是的，@Akina的查询更好：

SELECT id, company_id, image_path, ROW_NUMBER() OVER (PARTITION BY company_id) AS company_image_counter
FROM test_data2;

    id  company_id  image_path  company_image_counter  
------  ----------  ----------  -----------------------
     1         100  img_100_a                         1
     2         100  img_100_b                         2
     3         100  img_100_c                         3
     4         100  img_100_d                         4
     5         101  img_101_a                         1
     6         101  img_101_b                         2